Doi:10.1080/09540120500038058

AIDS Care, August 2005; 17(6): 661 Á/673 An empirical test of the Information, Motivation andBehavioral Skills model of antiretroviral therapyadherence K. R. AMICO1, J. TORO-ALFONSO2, & J. D. FISHER1 1Center for Health/HIV Intervention and Prevention, University of Connecticut, USA, and 2UniversityCenter for Psychological Services & Research, University of Puerto Rico, Puerto Rico AbstractNearly perfect adherence to demanding antiretroviral therapy (ART) is now recognized as essential forHIV-positive patients to realize its life sustaining benefits. Despite the dire consequences of non-adherence, a large number of patients do not follow their ART regimen. While many factors influenceadherence, the literature is dominated by studies on only one or a small set of them. Multivariate,theory-based models of adherence behavior are of great interest. The current study tested one suchmodel, the Information, Motivation and Behavioral Skills (IMB) model of ART adherence (Fisheret al., under review). A sample of HIV-positive patients on ART in clinical care in Puerto Rico(N 0/200) provided data on adherence-related information, motivation and behavioral skills as well asadherence behavior per se . Structural equation model tests used to assess the propositions of the IMBmodel of ART adherence provided support for the interrelations between the elements proposed bythe model and extended previous work. Implications for future research and intervention developmentare discussed.
In 1996, the introduction of ART for HIV provided one of the first opportunities to changewhat was often a fatal illness into a chronic, manageable disease. When taken as prescribed,ART has consistently been associated with decreases in viral load, increases in CD4 T-cellcounts and reduced rates of hospitalization, progression to AIDS and mortality (Arnstenet al., 2000; Bangsberg et al., 2000, 2004; Hogg, Yip, Chan, O’Shaughnessy, & Montaner,2000; Jensen-Fangel et al., 2004; Pradier et al., 2001; Roca, Gomez, & Arnedo, 2000).
Unfortunately, even the slightest digression from one’s prescribed regimen can make thesepotential benefits temporarily or permanently unattainable. ART adherence rates that fallbelow 80% have been associated with the development of antiretroviral drug resistantstrains of HIV (Bangsberg et al., 2000; Ickovics & Chesney, 1997; Knobel et al., 2002;Spire et al., 2002; Walsh, Horne, Dalton, Burgess, & Gazzard, 2001; Zaccarelli et al., 2002)that are often cross-resistant to ART medications falling in the same drug class, quicklyexhausting potential treatment options. Adherence rates below 90 Á/95% have beenassociated with rises in HIV RNA copy number (Blaschke, 1997; Knobel et al., 2002),significant decreases in CD4 counts (Singh et al., 1999) and increases in mortality rates Correspondence: Jeffrey D. Fisher, Center for Health/HIV Intervention and Prevention, 2006 Hillside Road, Unit1248, Storrs, CT 06269, USA. E-mail: [email protected] ISSN 0954-0121 print/ISSN 1360-0451 online # 2005 Taylor & Francis Group LtdDOI: 10.1080/09540120500038058 (Garcia de Olalla et al., 2002; Hogg et al., 2000). In sum, for HIV-positive individuals ableto maintain nearly perfect adherence, ART offers an opportunity for longer, healthier lives.
Failure to adhere may have dire consequences for the individual, and given the increasingnumbers of new HIV infections which involve treatment resistant strains of HIV (Boden etal., 1999; Hecht, Colfax, Swanson, & Chesney, 1998; Little et al., 2002; Richman, 2001),can also have a disastrous impact on the epidemic as a whole.
Despite the life-threatening consequences of non-adherence, about 57 Á/77% of indivi- duals on ART (Belzer, Fuchs, Luftman, & Tucker, 1999; Montessori et al., 2000; Rigsby etal., 2000; Singh et al., 1999; Spire et al., 2002) are unable to reach or maintain the 95%adherence rate now considered optimal (Bartlett, 2002; Low-Beer, Yip, O’Shaughnessy,Hogg, & Montaner, 2000; Paterson et al., 2000). This has created a critical need to identifyfactors influencing ART adherence and to develop interventions that target such factors.
Although substantial progress has been made in recent years in better understanding why individuals fail to adhere, well-tested, multivariate, conceptual models of ART adherenceare rare. Thus far, exploration of the determinants of ART adherence has been dominatedby the ‘single-variable’, predictive approach. While this work has provided valuableinformation regarding the associations between individual variables and adherence, itdoes not present a sufficiently complex view of the factors associated with ART adherence.
Multivariate, theory-based models are only now emerging (e.g., Fisher, Fisher, Amico, &Harman, under review), and the evaluation of such models is critical in the development ofeffective interventions. The current study was developed to evaluate one such model, IMB(Fisher et al., under review), which offers a dynamic, multivariate account of ARTadherence.
As indicated in Figure 1, the IMB model of adherence (Fisher et al., under review) identifies adherence-related information, motivation and behavioral skills as criticaldeterminants of ART adherence. Adherence-related information , a prerequisite of adequateadherence encompasses accurate information concerning one’s specific regimen, about howto utilize ART, about the requirements for adequate adherence, about specific side effectsassociated with one’s regimen, and information about potential drug interactions. Alsoincluded in the information construct are faulty heuristics and implicit theories (e.g., the others’ support for adherenceand motivation to complywith significant others’wishes Figure 1. IMB model of ART adherence (adapted from Fisher et al., under review).
notion that ‘If I’m feeling well, then missing a few doses doesn’t really matter’) play animportant role in negatively affecting adherence.
Adherence-related motivation refers to a personal and social motivation to follow one’s ART regimen as prescribed. Personal motivation includes one’s attitudes and beliefs aboutthe consequences outcomes of both optimal and sub-optimal adherence. Social motivationincludes perceptions of support for adherence behaviors from important others, as well asone’s motivation to comply with their wishes.
While the motivation and information constructs of the model may share some variability, motivation to adhere to one’s ART regimen is not necessarily related to one’slevel of accurate information about one’s regimen. As indicated in Figure 1, one may behighly motivated to adhere to ART, yet have inaccurate information about ART. Similarly,one may have entirely accurate information about ART, but have little motivation to adhere.
Adherence-related information and motivation are each related to the performance of ART adherence behavioral skills , specifically when individuals are well informed about ARTand motivated they acquire and execute behaviors for adequately and consistently ensuringART adherence. Adherence-related behavioral skills include both the objective ability andperceived efficacy (Bandura, 1977) for performing critical behaviors, such as acquiring andself-administering ART medications consistently over time, achieving a good fit betweenone’s regimen and the natural ecology of daily life, taking steps to minimize side effects,obtaining ART-related information and/or support when needed, and developing strategiesto reward and reinforce ART adherence behaviors.
As indicated in Figure 1, for complex or novel behaviors the IMB model posits a direct relation between behavioral skills and adherence behavior, in which adherence-relatedinformation and motivation relate to adherence behavior primarily through behavioral skills.
In effect, since the skills presently required to establish and maintain adequate ARTadherence are typically complex or novel, behavioral skills will mediate the relationsbetween information and motivation and ART adherence behavior, and the direct pathsfrom information and motivation to adherence behavior are anticipated to be negligible.
Thus, an informed and motivated person who lacks the objective skills necessary to acquire,self-cue, and take medication or the confidence in his or her efficacy to do so, will havedifficulty establishing or maintaining adequate adherence. If at some point regimens areavailable for which the requisite skills for ART adherence are simple or automated (e.g., atransdermal patch that must be changed monthly), information and motivation would beexpected to relate to adherence behavior directly.
Finally, the IMB model of ART adherence (Fisher et al., under review) predicts that high levels of ART adherence will result in favorable health outcomes, and that poor adherencewill result in unfavorable health outcomes, which is quite consistent with the currentliterature (e.g., Hogg et al., 2000; Pradier et al., 2001; Roca et al., 2000). Moreover,longitudinally, these favorable or unfavorable health outcomes will affect subsequent levelsof adherence-related information, motivation, and behavioral skills through a feedback loop(for details, see Fisher et al., under review). The model also identifies several potentialpersonal or environmental factors (e.g., substance use problems, unstable housing, orsevere mental health issues) that will moderate (i.e., strengthen or weaken) the relationsbetween adherence-related information, motivation, behavioral skills, and adherence. Thus,the IMB model of adherence provides a comprehensive, multivariate, theory-groundedapproach to conceptualizing ART adherence behavior.
Recently, Starace, Massa, Mariniello, Amico, & Fisher (in press) presented an initial multivariate test of the IMB model of ART adherence with a sample of HIV-positive patients in clinical care in Italy. Their results generally supported the model, but hadlimitations in terms of small sample size. Moreover, the results were limited to a singlegeographic locality and it was unclear if their findings would generalize to other countriesand/or cultures, or to ART adherence in the context of other medical health care systems.
The current study tests the IMB model of ART adherence with a larger sample than Staraceet al. (in press) in a sample of HIV-positive patients receiving clinical care in San Juan,Puerto Rico. Specifically, the determinants of ART adherence identified by the IMB modelof ART adherence (adherence-related information, motivation and behavioral skills) wereused in a series of structural equation model tests to evaluate both the fit of the IMB modelof adherence to the sample data and whether or not the structural assumptions of themodel (e.g., mediation) appropriately described the interrelations among the IMBmodel constructs.
HIV-positive patients in clinical care in four HIV care clinics in and near San Juan, PuertoRico (N 0/200) completed measures of IMB model constructs with the assistance ofinterviewers between March and May 2001. Four trained interviewers recruited partici-pants from the waiting areas of the clinic sites and administered questionnaires. Three ofthe clinics serviced only HIV-positive patients during the days of recruitment and oneserved patients with diverse medical concerns. The majority of participants came from theHIV specialty clinics. For the clinics where all patients were HIV-positive, an announce-ment regarding the project and a call for participants, as well as written material placedaround the waiting area detailing the project, were used to recruit patients. Case managersalso referred patients for participation in the project. Participants from the clinic that serveddiverse medical needs were recruited discretely by referral from staff. Only those who werecurrently prescribed antiretroviral medications were eligible to participate. Whether or not apatient was currently on ART was confirmed by the clinic staff prior to participation.
Participants were reimbursed $15.00 for completion of the survey. All interested andeligible patients signed informed consents prior to participation in the research. All of theseresearch procedures were approved by relevant university and hospital/clinic institutionalreview boards.
The Spanish version of the IMB ART adherence questionnaire (Amico et al., 2001)contained sets of ART adherence information, motivation, behavioral skills and behavioritems consistent with those used in previous research (Starace et al., in press). In addition, amodified version of the Adherence to Combination Therapy Questionnaire (AACTG)(Chesney et al., 2000) was used to assess rates of ART adherence. The original AACTGuses self-report of doses missed over the past three days. Because we were concerned thatpatients in the current population might be reluctant to report actually missing doses, wemodified the instrument and asked participants to report number of doses taken .
Adherence was then calculated as the number of doses taken over the number prescribedduring the time period. Participants completed these measures in an interview format.
Responses from the AACTG questionnaire were used to calculate adherence over the past three days. Rates of adherence were defined as the percentage of prescribed pillsactually taken of the total number of pills prescribed over a three-day period.
Adherence-related information was measured by three behaviorally relevant knowledge items that could be rated by the participant as ‘true’, ‘false’ or ‘not sure/don’t know’. Thesemapped closely onto the information construct of the IMB model (Figure 1) and onto theitems used in Starace et al. (in press). Responses to the three items ‘I know what to do’ (e.g.,‘Whether or not to take the pill later if I miss a dose of any of my HIV medications’,‘If I take non-prescription drugs, e.g., Tylenol or recreational drugs such as crack, cocaine,heroine, etc., I know how they could affect the way my HIV pills work’ and ‘I know whatthe side effects of my combination therapy medication might be’) were scored as follows:three 0/correct response, two 0/unsure and one 0/incorrect response. Total scores couldrange from three to nine points.
Similar to Starace et al. (in press), we then created two groups to represent participants who were very well informed and those who were not as well informed. Since the median inthe sample fell between eight and nine points, we defined those who were well informed ashaving perfect scores (nine), and those who were less well informed as having less thanperfect scores. One hundred and thirty-eight participants (69%) were classified as wellinformed, while 62 (31%) were classified as less well informed. These two groups can beunderstood as reflecting a participant’s relative amount of accurate information in thecontext of a generally well-informed cohort.
Adherence-related motivation was also measured consistent with the IMB model (Figure 1) and with Starace et al.’s (in press) measure of this construct. Participants used a five-point Likert scale to respond to each of eight items designed to tap aspects of adherence-related motivation, including ‘attitudes about ART and ART adherence’ (e.g., ‘Continuingto take my medications as directed when I am experiencing side effects’ would be rated verygood to very bad) and ‘perceptions of significant others’ support for ART adherence’ (e.g.,‘The people who are important to me think I should take all my combination therapymedications according to the doctor’s orders’ would be rated strongly disagree to stronglyagree). Item responses were summed and averaged. As we anticipated, given that themotivation construct was comprised of items that may not co-vary (e.g., having a positiveattitude towards ART medications can be independent of the degree of support one has formedication-taking behavior from those who are important to one), the scale’s inter-itemconsistency was low (a 0/0.52, 0.62 standardized) and no single item deletion offered animprovement in consistency. Total motivation scores represented the average responseacross the eight items, and ranged from one to five, where five represents higher motivation.
Across the sample, participants demonstrated a high degree of adherence motivation(M 0/4.72, SD 0/0.262).
Finally, consistent with the IMB model and with Starace et al. (in press), adherence- related behavioral skills were represented by six behaviorally relevant items that representedthe diverse set of skills required for adequate ART adherence. Items included assessments ofthe participants’ skills for taking medication as prescribed in the presence of a number ofbarriers (e.g., ‘I have no problems taking my combination therapy medications correctly,even when it’s difficult to work them into my schedule’, ‘I can always take my combinationtherapy medications according to the doctor’s orders (even when I am at work, when I amout with my friends or during the week-end)’ and ‘If I need more information about any ofthe combination therapy drugs I’m taking, I feel confident that I know how to reachsomeone to find out the information’). Participants rated the extent to which they agreed or disagreed with each statement on a five-point scale, and scores were recoded so that higherscores reflected greater skills. As might be expected given the diverse barriers representedby behavioral skills items, inter-item consistency was low (a 0/0.35, 0.41 standardized).
Scores a composite skills index. Overall, the sample reported a high level of behavioral skills(M 0/4.40, SD 0/0.621).
Evaluation of the IMB model of ART adherence included several different analyses; the firstinvolved the evaluation of a full IMB model, where the three IMB model constructs(information, motivation and behavioral skills) were used in a structural model to accountfor optimal adherence. This analysis focused on assessing the structural mediationhypotheses of the model: that information and motivation would significantly relate tobehavioral skills and not directly to adherence behavior and that behavioral skills wouldsignificantly relate to optimal adherence. As a just-identified model, parameter estimateswere the primary tools for evaluation.
The second analysis allowed for an assessment of the IMB model constructs’ association with optimal adherence. We fitted a restricted model where each of the paths betweeninformation and adherence behavior and between motivation and adherence behavior wereremoved from the model. Standard model fit indices (e.g., x2, CFI, RMSEA: Bollen, 1989;Kline, 1998) were used to evaluate this restricted model, providing an evaluation of thegeneral fit of the critical IMB constructs to the sample data.
The third analysis focused on the structural features of the IMB model of ART adherence by assessing the differences in the x2 values obtained when the less restricted models werecompared to the fit obtained for the restricted (mediated) model. All path analyses wereestimated using MPLUS software (Muthen & Muthen, 2001), with weighted least squaremean and variance adjusted chi-square (WLSMV) estimation (Muthen & Muthen, 2002)to assess the parameters in the full model and fit to the sample data in the restricted model.
The comparative fit between the restricted model and alternative models were evaluatedusing standard weighted least square (WLS) estimation and associated x2 difference tests(Muthen & Muthen, 2002).
The majority of the 200 HIV-positive participants (99%) attended one of the three HIVspecialty care clinics involved in the study. Sixty-five percent of the sample was male, 34%was female and 0.5% was transgender. The majority reported being Latino/a (84%),followed by Black (7%) and White (5%). The age of participants ranged from 19 Á/81 years,with an average of 39 (SD 0/8.93). Thirteen percent had earned a bachelor’s degree, 23%had some university training, 25% had earned a high-school diploma and 21% hadcompleted some high school. The majority of participants were unemployed (59%), hadgovernment provided/funded insurance (86%) and lived on marginal incomes (84%reported annual family incomes of US$10,000 or less). Most were heterosexual (73%),with 19% reporting homosexual and 8% reporting bisexual orientations. The averagenumber of years since HIV diagnosis was 6.82 (SD 0/4.53; range 0/0.08 to 21 years). Forty-two percent reported contracting HIV from heterosexual sex, 21% through sex with men, 34% through needle/equipment sharing, 1% through blood transfusion and 2% reported‘other’. This distribution differs slightly from the epidemiological trend of the epidemic inPuerto Rico, in that the current sample has lower rates of IDU-infected individuals (34% inthe current sample versus 51% in Puerto Rico according to 2001 estimates) (CDC, 2004).
Complete AACTG data was provided by 196 participants. Participants’ ART regimensincluded one to six antiretroviral medications (M 0/2.61, median 0/3, SD 0/0.843), with atotal of one to 11 doses per day (M 0/4.91, median 0/5, SD 0/1.57) and a total number ofpills per day ranging from one to 11 (M 0/5.26, median 0/6, SD 0/2.05). The most commonantiretroviral in the sample’s ART regimens was combivir, followed by viracept and zerit.
Participants were on ART for an average of 3.78 years (SD 0/2.85). Adherence, asmeasured by the AACTG, ranged from 0 Á/100%, with an average of 91% adherence.
The distribution of percent adherence was highly kurtotic (9.56) and negatively skewed( (/3.15). Rates of adherence were used to create a dichotomous variable reflecting whetheror not a participant met the currently recognized standard of ]/95% (Bartlett, 2002; Low-Beer et al., 2000) for adequate adherence. One hundred and fifty-six participants (80%)reported optimal adherence (95% or greater rates of adherence) and 40 participants (20%)reported sub-optimal adherence (less than 95% rates of adherence). This dichotomousvariable, with optimal adherence scored as one and sub-optimal adherence scored as zero,served as the outcome variable in our analyses.
Evaluation of the IMB model of ART adherence Prior to evaluating the IMB model of ART adherence, we assessed the general equivalenceof participants between sites. Although four clinics agreed to participate in the currentresearch, we used the data from three clinics (N 0/48, 88 and 62) because only two surveyswere collected at the fourth clinic site, which was the only participating site that was aprivate general practitioner’s office. Because this office treated a small number of HIV-positive patients, privacy issues precluded aggressive or overt recruitment strategies at thissite. The included clinic sites were comparable in terms of years patients were on ART, ratesof adherence, gender distributions, reported sexual orientation, level of educationalattainment and how the patient contracted HIV. Although we did find differences in termsof age and annual income, these were not included as covariates as they demonstrated nosignificant relation to adherence. Across the demographic variables, and relative to thedependent variable of interest, sites appeared generally comparable and the data collectedfrom each site was combined for analyses.
Using the full set of data, collapsing across clinic sites, the IMB model of ART adherence was assessed as a full model using four variables: adherence-related information,motivation, behavioral skills and adherence per se (defined as optimal versus sub-optimaladherence). The full path model was estimated using weighted least square mean andvariance adjusted chi-square (WLSMV) (Muthen & Muthen, 2002) and parameterestimates were evaluated in terms of significance and direction. As depicted in Figure 2,the mediated structural hypotheses of the IMB model of ART adherence were supported.
Specifically, both adherence-related information and motivation related significantly andpositively to behavioral skills, but did not significantly relate directly to optimal adherence.
Moreover, as predicted by the model, adherence-related behavioral skills were significantlyand positively associated with optional adherence. The elements in the model accounted for Figure 2. IMB model of ART adherence: Full model. WLSMV estimation.
*p B/0.05.; N 0/200; acoefficient for the path from the dichotomous information variable to the continuousbehavioral skills variable (0.135) was divided by the standard deviation of the dichotomous information deficitvariable 0.4636 to produce the standardized coefficient (0.291).
19% of the variability in adherence group membership. Thus, support was provided for theIMB model’s major assertions that adherence-related information, motivation andbehavioral skills are important determinants of adherence behavior and that ART adherencebehavior is influenced by information and motivation primarily through their effects onbehavioral skills.
In order to assess the overall fit of a mediated IMB model of ART adherence, we evaluated a mediated (restricted) model’s fit to the sample data, again using WLSMVestimation. As depicted in Figure 3, the restricted model demonstrated good fit to thesample data (x2 (2, N 0/200) 0/5.063, p 0/0.078, CFI 0/0.958, RMSEA 0/0.088), account-ing for 17% of the variability in adherence group membership. This restricted model wasthen used in comparison to alternative models that added a non-mediated path frominformation to optimal adherence and one that added a non-mediated path from motivationto optimal adherence to test for mediation. Mediation would be indicated by non-significant differences in model fit between the restricted (mediated) model and alternativemodels that are less restrictive.
Figure 3. IMB model of ART adherence: Restricted model. WLSMV estimation. x2 (2, N 0/200) 0/5.063,p 0/0.078, CFI 0/0.958, RMSEA 0/0.088.
*p B/0.05; N 0/200; acoefficient for the path from the dichotomous information variable to the continuousbehavioral skills variable (0.164) was divided by the standard deviation of the dichotomous information deficitvariable 0.4636 to produce the standardized coefficient (0.354).
In order to perform chi-square difference tests, the restricted model was first re-estimated using the WLS estimator, as WLSMV chi-square cannot be compared across models(Muthen & Muthen, 2002). The WLS generated model fit for model (x2 (2, N 0/200) 0/4.299) was compared to the fit of the first alternative modelwhere the non-mediated path from information to optimal adherence was added (x2 (1,N 0/200) 0/1.760). The resulting difference in fit (x2 (1, N 0/200) 0/2.539, p 0/0.11)suggested that the two models were generally comparable and that the more parsimoniousmodel provided a sufficient and preferable fit to sample data. Similarly, the addition of apath from motivation to optimal adherence (x2 (1, N 0/200) 0/3.320) did not significantlyimprove model fit (x2 (1, N 0/200) 0/0.979, p 0/0.32). Thus, it appeared that the mediatedmodel provided a good fit to the sample data.
The current study contributes to growing efforts to understand ART adherence behavior inthe context of theoretical models of health behavior. The IMB model of ART adherence(Fisher et al., under review) is one of the few models of ART adherence in the literature thatis based on a well-validated theory of health behavior change (Fisher, Fisher, & Harman,2003) and clearly articulates the personal, social and environmental factors presumed toinfluence ART adherence, as well as the manner in which such factors interrelate. It gainedpreliminary support with a sample of HIV-positive patients receiving care in Italy (Staraceet al., in press) and the current study extended that support in important ways. With alarger sample, in the context of a very different health care system, with somewhat differentbut parallel measures and in a different country and culture, we assessed the IMB model ina sample of HIV-positive patients receiving clinical care in Puerto Rico.
The IMB model of ART adherence (see Figure 1) hypothesizes that adherence-related information, motivation and behavioral skills are the critical determinants of adherencebehavior. Accurate information about one’s ART regimen and about what constitutesadequate adherence, high levels of adherence-related motivation and strong adherence-related behavioral skills are posited to underlie optimal adherence. The model also specifiesthat the relations between adherence-related information and motivation and adherencebehavior are mediated by adherence-related behavioral skills. Simply stated, even a wellinformed or highly motivated individual will have difficulty achieving and sustainingoptimal adherence if he or she lacks the objective skills required to acquire or self-administer medication or feels incapable of performing such behaviors.
Results of the current study provided strong support for the IMB model of ART adherence. The IMB model of ART adherence provided a good fit to the sample data(see Figure 2). Evaluation of a full, non-mediated model indicated that information andmotivation were positively associated with adherence-related behavioral skills, and thatbehavioral skills were in turn positively related to optimal adherence. The path coefficientsfor the non-mediated or direct relations between adherence-related information andoptimal adherence and between adherence-related motivation and optimal adherence werenon-significant. In comparison to partially mediated models, the fully mediated modelprovided a more parsimonious explanation of adherence behavior. Thus, the analysesprovided support for both the overall and specific structural propositions of the IMB model.
The high rates of adherence we found in our sample of HIV-positive patients in clinical care in Puerto Rico stand in contrast to previous research in the United States, whichgenerally has found that between 57% and 77% of HIV- positive patients report adherence rates below 90% (Belzer et al., 1999; Montessori et al., 2000; Rigsby et al., 2000; Singh etal., 1999; Spire et al., 2002), but is quite consistent with the higher rates of adherencereported in resource poor areas and countries (Attawell & Mundy, 2003). For example,Oyugi et al. (2004) recently reported average adherence rates, across a number of measures,of 91 Á/94% adherence in their work with resource poor HIV-positive persons in Uganda.
Additionally, Attawell and Mundy’s (2003) recent review summarizes a number of studiesthat have found similarly high rates of adherence in resource poor countries. Thus, in areaswhere resources are limited and there are substantial barriers to even gaining access to ART,perhaps non-adherence may be less prevalent. While we recognize that somewhat higherrates of adherence have been associated with self-report measures of adherence, incomparison to more objective measures (Arnsten et al., 2001; Walsh, Mandalia, & Gazzard,2002), the validity of self-report in terms of association with disease progression and bloodconcentrations of ART medications (Haubrich et al., 1999; Hecht et al., 1998; Kleebergeret al., 2001; Knobel et al., 2002; Moatti & Spire, 2000; Murri et al., 2000; Nieuwkerk et al.,2001) has been well established. Thus, the rates of adherence reported by the currentsample may reflect a genuine high degree of adherence in HIV-positive individuals receivingcare in Puerto Rico, but further research using a multiple methods approach to assessadherence behavior is needed.
Similar to previous research with the IMB model in other health domains (Fisher & Fisher, 1993; Misovich, Fisher, & Fisher, 1998), the present assessment of adherence-related behavioral skills, which conceptually includes both objective skills and perceivedefficacy, focused operationally primarily on participants’ perceived efficacy in performingcomplex adherence behaviors over time. Assessing objective skills related to adherence aswell would require the repeated observations of patients as they navigate in their naturalenvironments, which was not possible in the current study. While a complete assessment ofbehavioral skills would include perceptions of efficacy and objective skills, it is notuncommon for behavioral skills to be assessed primarily through perceived efficacy due tothe difficulty associated with assessment of the objective skills. Nevertheless, it has beenshown that perceived efficacy is related to levels of objective behavioral skills (Williams etal., 1998). Thus, while perceived efficacy and behavioral skills represent conceptuallydistinct constructs, they have, in practice, shown a positive relation in studies of healthbehavior (King, Humen, Smith, Phan, & Teo, 2001; Stewart, Strack, & Graves, 1999).
In the current research, the assessment of adherence did not distinguish between different types of regimen. While this is consistent with previous literature, recent work (Bangsberg etal., 2003) has suggested that the relation between rates of adherence and outcomes such asoptimal viral suppression and the development of resistance are complex and partiallydependent on the specific characteristics of one’s regimen (e.g., boosted versus non-boostedPI regimens). In the current sample, 97% of the participants were on non-boostedregimens, which do appear to demonstrate the bell curve association between thedevelopment of resistance and rates of adherence (Bangsberg, Moss, & Deeks, 2004). Asboosted regimens become more common, specific regimen type should be taken intoaccount, as the adherence-resistance relationship may differ for boosted versus non-boostedregimens.
Seven years after the introduction of ART, the study of adherence to its demanding and often complicated regimens has made substantial progress. The development of theories ofART adherence that draw from well-validated models of health behavior and behaviorchange is a promising new direction that provides a structure to the current literature andease of translation from theory to practice. The identification of critical determinants of ART adherence and how these factors might interact to influence behavior is an essentialprerequisite in helping HIV-positive patients to achieve and maintain high levels of ARTadherence. The current study offers support for the IMB model of adherence and furthersefforts to better understand ART adherence as a dynamic behavior that occurs in thecontext of an individual’s personal, social and environmental context. Ultimately, the truepotential of any theory-grounded model of ART adherence will involve the evaluation of itsability to produce effective ART adherence interventions. Such efforts with the IMB modelof ART adherence are underway and will make substantial contributions to the theory-driven efforts to prolong and improve the quality of life for HIV-positive patients on ARTthrough behavior change interventions.
The work in this article was partially supported by NIMH grants 5R01MH59473-02 andR01MH59473.
We thank Ivan Andujar and Jamie Calderon from the University of Puerto Rico for their assistance in collecting the data presented and managing the data collection process.
We also thank Jennifer Harman and Angela Bryan for their assistance in analyzing the data.
Amico, K. R., Toro-Alfonso, J., Fisher, J. D., Andujar, I., & Acaldero´n, J. (2001). Adaptation of the adherence to combination therapy questionnaire and interview. Unpublished manuscript , University of Puerto Rico.
Arnsten, J. H., Demas, P. A., Farsadegan, H., Grant, R. W., Gourevitch, M. N., Chang, C. J., et al. (2001).
Antiretroviral therapy adherence and viral suppression in HIV-infected drug users: Comparison of self-reportand electronic monitoring. Clinical Infections Disease , 33 , 1417 Á/1423.
Arnsten, J., Demas, P., Gourevitch, M., Buono, D., Farzadegan H., & Schoenbaum, E. (2000). Adherence and viral load in HIV-infected drug users: Comparison of self-report and medication event monitors (MEMS).
Paper presented at the Seventh Conference on Retroviruses and Opportunistic Infections, San Francisco.
Attawell, K., & Mundy, J. (2003). Provision of antiretroviral therapy in resource-limited settings: A review of experience up to August 2003. The health systems resource centre (HSRC). Retrieved on August 26 2004 fromhttp://www.who.int/3by5/publications/documents/en/ARTpaper_DFID_WHO.pdf Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84 , Bangsberg, D. R., Charlebois, E. D., Grant, R. M., Holodniy, M., Deeks, S. G., Perry, S. F., et al. (2003). High levels of adherence do not prevent accumulation of HIV drug resistance. AIDS , 17 , 1925 Á/1932.
Bangsberg, D. R., Hecht, F. M., Clague, H., Charlebois, E., Ciccarone, D., Chesney, M., & Moss, A. R. (2000).
Provider estimate and structured patient report of adherence compared with unannounced pill count. Paperpresented at the Seventh Conference on Retroviruses and Opportunistic Infections, San Francisco.
Bangsberg, D. R., Moss, A. R., & Deeks, S. G. (2004). Paradoxes of adherence and drug resistance to HIV antiretroviral therapy. Journal of Antimicrobial Chemotherapy, 53 , 696 Á/699.
Bartlett, J. A. (2002). Addressing the challenges of adherence. Journal of Acquired Immune Deficiency Syndromes , 29 , Belzer, M. E., Fuchs, D. N., Luftman, G. S., & Tucker, D. J. (1999). Antiretroviral adherence issues among HIV- positive adolescents and young adults. Journal of Adolescent Health , 25 , 316 Á/319.
Blaschke, T. (1997). Noncompliance and resistance to protease inhibitors. Paper presented at the Fourth Conference of Retroviruses and Opportunistic Infections, Washington, DC.
Boden, D., Hurley, A., Zhang, L., Cao, Y., Guo, Y., Jones, E., et al. (1999). HIV-1 drug resistance in newly infected individuals. Journal of the American Medical Association , 282 , 1135 Á/1141.
Bollen, K. A. (1989). Structural equations with latent variables . New York: Wiley CDC (2004). AIDS Public Information Data Set (APIDS) 2001. Retrieved July 7, 2004 from http://www.cdc.gov/ Chesney, M. A., Ickovics, J. R., Chambrers, 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 AdherenceInstruments. AIDS Care , 12 , 255 Á/266.
Fisher, J. D., Fisher, W. A., Amico, K. R., & Harman, J. J. (under review). An information, motivation, behavioral, skills model of adherence to antiretroviral therapy.
Fisher, W. A., & Fisher, J. D. (1993). A general social psychological model for changing AIDS risk behavior. In J. B.
Pryor (Ed.), The social psychology of HIV infection (pp. 127 Á/153). Hillsdale, NJ: Lawrence Erlbaum.
Fisher, W. A., Fisher, J. D., & Harman, J. J. (2003). The information, motivation, behavioral, skills model as a general model of health behavior change: Theoretical approaches to individual-level change. In J. Suls, & K.
Wallston (Eds.), Social psychological foundations of health (pp. 127 Á/153). London: Blackwell.
Garcia de Olalla, P., Knobel, H., Carmona, A., Guelar, A., Lopez-Colomes, J. L., & Cayla, J. A. (2002). Impact of adherence and highly active antiretroviral therapy on survival in HIV-infected patients. Journal of AcquiredImmune Deficiency Syndrome , 1 , 105 Á/110.
Haubrich, R. H., Little, S. J., Currier, J. S., Forthal, D. N., Kemper, C. A., Beall, G., et al. (1999). The value of patient-reported adherence to antiretroviral therapy in predicting virologic and immunologic response. AIDS ,13 , 1099 Á/1107.
Hecht, F. M., Colfax, G., Swanson, M., & Chesney, M. A. (1998). Adherence and effectiveness of protease inhibitors in clinical practice. Abstract of presentation from the Fifth conference on Retroviruses andOpportunistic Infections, Chicago.
Hogg, R. S., Yip, B., Chan, K., O’Shaughnessy, M. V., & Montaner, S. G. (2000). Nonadherence to triple combination therapy is predictive of AIDS progression and death in HIV-positive men and women. Paperpresented at the Seventh Conference on Retroviruses and Opportunistic Infections, San Francisco.
Ickovics, J., & Chesney, M. (1997). Issues regarding antiretroviral treatment for patients with HIV-1 infection.
Journal of the American Medical Association , 278 , 1233 Á/1234.
Jensen-Fangel, A., Pederson, L., Pederson, C., Larsen, C. S., Tauris, P., Moller, A., et al. (2004). Low mortality in HIV-infected patients starting highly active antiretroviral therapy: A comparison with the general population.
AIDS , 18 , 89 Á/97.
King, K. M., Humen, D. P., Smith, H. L., Phan, C. L., & Teo, K. K. (2001). Psychosocial components of cardiac recovery and rehabilitation attendance. Health , 85 , 290 Á/294.
Kleeberger, C. A., Phair, J. P., Strathdee, S. A., Detels, R., Kingsly, L., & Jacobson, L. P. (2001). Effect of computer-assisted self-interviews on reporting of sexual HIV risk behaviors in a general populations sample:A methodological experiment. Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology, 26 ,82 Á/92.
Kline, R. B. (1998). Principles and practice of structural equation modelling . New York: Guilford Press.
Knobel, H., Alonso, J., Casado, J. L., Collazos, J., Gonzalez, J., Ruiz, I., et al. (2002). Validation of a simplified medication adherence questionnaire in a large cohort of HIV-infected patients: The GEEMA Study. AIDS , 16 ,605 Á/613.
Little, S. J., Holte, S., Routy, J., Daar, E. S., Markowitz, M., Collier, A. C., et al. (2002). Antiretroviral-drug resistance among patients recently infected with HIV. New England Journal of Medicine , 347 , 385 Á/394.
Low-Beer, S., Yip, B., O’Shaughnessy, M., Hogg, R., & Montaner, J. (2000). Adherence to triple therapy and viral load response. Journal of Acquired Immune Deficiency Syndromes , 23 , 360 Á/361.
Misovich, S. J., Fisher, W. A., & Fisher, J. D. (1998). A measure of AIDS-prevention information, motivation, behavioral skills and behavior. In C. Davis, et al. (Eds.), Sexuality related measures: A compendium (2nd edition).
Thousand Oaks, CA: Sage; New York: Guilford Press.
Moatti, J. P., & Spire, B. (2000). Living with HIV/AIDS and adherence to antiretroviral treatments. In J. P. Moatti, Y. Souteyrand, A. Prieur, T. Sandfort, & P. Aggleton (Eds.), AIDS in Europe: New challenges for the social sciences(pp. 57 Á/73). New York: Routledge.
Montessori, V., Heath, K. V., Yip, B., Hogg, R. S., O’Shaughnessy, M. V., & Montaner, S. G. (2000). Predictors of adherence with triple-combination antiretroviral therapy. Paper presented at the Seventh Conference onRetroviruses and Opportunistic Infections, San Francisco.
Murri, R., Ammassari, A., Gallicano, K., De Luca, A., Cingolani, A., Jacobson, D., et al. (2000). Patient-reported nonadherence to HAART is related to protease inhibitor levels. Journal of Acquired Immune Deficiency Syndromes ,24 , 123 Á/128.
Muthen, L. K., & Muthen, B. O. (2001). MPLUS; Statistical analysis with latent variables user’s guide (2nd edition).
Muthen, L. K., & Muthen, B. O. (2002). Modeling with categorical latent variables using MPLUS. Course offered at MPLUS Short Courses in Washington, DC.
Nieuwkerk, P. T., Sprangers, M. A., Burger, D. M., Hoetelmans, R. M., Hugen, P. W., Danner, S. A., et al. (2001).
Limited patient adherence to highly active antiretroviral therapy for HIV-1 infection in an observational cohortstudy. Archives of Internal Medicine , 161 , 1962 Á/1968.
Oyugi, J. H., Byakika-Tusiime, J., Pharm, B., Charlebois, E. D., Kityo, C., Mugerwa, R., et al. (2004). Multiple validated measures of adherence indicates high levels of adherence to generic HIV antiretroviral therapy in aresource limited setting. Journal of Acquired Immune Deficiency Syndromes , 36 , 1100 Á/1102.
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 , 21 Á/30.
Pradier, C., Carrieri, P., Bentz, L., Spire, B., Dellamonica, P., & Moreau, J. P. (2001). Impact of short-term adherence on virological and immunological success of HAART: A case study among French HIV-infectedIDUs. International Journal of STD & AIDS , 12 , 324 Á/328.
Richman, D. D. (2001). Clinical implications of HIV fitness. Paper presented at the 41st Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago.
Rigsby, M. O., Rosen, M. I., Beauvais, J. E., Cramer, J. A., Rainey, P. M., O’Malley, S. S., et al. (2000). Cue-dose training with monetary reinforcement: Pilot study of an antiretroviral adherence intervention. Journal of GeneralInternal Medicine , 15 , 841 Á/847.
Roca, B., Gomez, C. J., & Arnedo, A. (2000). Adherence, side effects and efficacy of stavudine plus lamivudine plus nelfinavir in treatment-experienced HIV-infected patients. Journal of Infection , 41 , 50 Á/54.
Singh, N., Berman, S. M., Swindells, S., Justis, J. C., Mohr, J. A., Squier, C., et al. (1999). Adherence of human immunodeficiency virus-infected patients to antiretroviral therapy. Clinical Infectious Diseases , 29 , 824 Á/830.
Spire, B., Duran, S., Souville, M., Leport, C., Raffi, F., Moatti, J. P., et al. (2002). Adherence to highly active antiretroviral therapies (HAART) in HIV-infected patients: From a predictive to dynamic approach. SocialScience and Medicine , 54 , 1481 Á/1496.
Starace, F., Massa, A., Mariniello, A. P. S., Amico, K. R., & Fisher, J. D. (in press). Application of the IMB model to assess HIV treatment adherence. Journal of Health Psychology (in press).
Stewart, J., Strack, S., & Graves, P. (1999). Self-efficacy, outcome expectancy, dental health value and dental plaque. American Journal of Health Behavior , 23 (4), 303 Á/310.
Walsh, J. C., Horne, R., Dalton, M., Burgess, A. P., & Gazzard, B. G. (2001). Reasons for nonadherence to antiretroviral therapy: Patients’ perspectives provide evidence of multiple causes. AIDS Care , 13 , 709 Á/720.
Walsh, J. C., Mandalia, S., & Gazzard, B. G. (2002). Responses to a one-month self-report on adherence to antiretroviral therapy are consistent with electronic data and virological treatment outcome. AIDS , 16 ,269 Á/277.
Williams, S. S., Doyle, T. M., Pittman, L. D., Weiss, L. H., Fisher, J. D., & Fisher, W. A. (1998). Role-played safer sex skills of heterosexual college students influenced by both personal and partner factors. AIDS and Behavior ,2 , 177 Á/187.
Zaccarelli, M., Barracchini, A., De Longis, P., Perno, C. F., Soldani, F., Liuzzi, G., et al. (2002). Factors related to virologic failure among HIV-positive injecting drug users treated with combination antiretroviral therapyincluding two nucleoside reverse transcriptase inhibitors and nevirapine. AIDS Patient Care and STDs , 16 ,67 Á/73.

Source: http://jtoro.uprrp.edu/Articulos%201/An%20empirical%20test.pdf