Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Patient and Regimen Characteristics Associated with Self-Reported Nonadherence to Antiretroviral Therapy

  • Patrick S. Sullivan ,

    To whom correspondence should be addressed. E-mail:

    Affiliation Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America

  • Michael L. Campsmith,

    Affiliation Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America

  • Glenn V. Nakamura,

    Affiliation Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America

  • Elin B. Begley,

    Affiliation Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America

  • Jeffrey Schulden,

    Affiliation Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America

  • Allyn K. Nakashima

    Affiliation Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America

Patient and Regimen Characteristics Associated with Self-Reported Nonadherence to Antiretroviral Therapy

  • Patrick S. Sullivan, 
  • Michael L. Campsmith, 
  • Glenn V. Nakamura, 
  • Elin B. Begley, 
  • Jeffrey Schulden, 
  • Allyn K. Nakashima



Nonadherence to antiretroviral therapy (ARVT) is an important behavioral determinant of the success of ARVT. Nonadherence may lead to virological failure, and increases the risk of development of drug resistance. Understanding the prevalence of nonadherence and associated factors is important to inform secondary HIV prevention efforts.

Methodology/Principal Findings

We used data from a cross-sectional interview study of persons with HIV conducted in 18 U.S. states from 2000–2004. We calculated the proportion of nonadherent respondents (took <95% of prescribed doses in the past 48 hours), and the proportion of doses missed. We used multivariate logistic regression to describe factors associated with nonadherence. Nine hundred and fifty-eight (16%) of 5,887 respondents reported nonadherence. Nonadherence was significantly (p<0.05) associated with black race and Hispanic ethnicity; age <40 years; alcohol or crack use in the prior 12 months; being prescribed ≥4 medications; living in a shelter or on the street; and feeling “blue” ≥14 of the past 30 days. We found weaker associations with having both male-male sex and injection drug use risks for HIV acquisition; being prescribed ARVT for ≥21 months; and being prescribed a protease inhibitor (PI)-based regimen not boosted with ritonavir. The median proportion of doses missed was 50%. The most common reasons for missing doses were forgetting and side effects.


Self-reported recent nonadherence was high in our study. Our data support increased emphasis on adherence in clinical settings, and additional research on how providers and patients can overcome barriers to adherence.


The full benefit of antiretroviral therapy (ARVT) in HIV-infected patients requires patients to be prescribed appropriate therapy, the virus to be sensitive to the prescribed therapy, and patients to adhere to the prescribed regimen [1]. Evidence from clinical trials suggests that very high levels of adherence (>95%) are required for optimally suppressing viral load and preventing the emergence of resistant virus [2], [3]. Even among those with low CD4 counts at the initiation of ARVT, adherence is a strong predictor of increase in CD4 count after initiation of combination ARVT [4]. There may also be significant public health benefits associated with prescription of antiretroviral therapy and good adherence: on a population basis, lower viral loads are predicted to be associated with decreased sexual [5], perinatal [6], and injection-related [7] transmission of HIV.

Therefore, understanding the prevalence of nonadherence, its associated factors, and reasons for nonadherence is an important clinical and public health goal. Behaviors, such as adherence, that are related to HIV disease progression are measured as part of behavioral surveillance projects conducted in the United States [8]. Generally, behavioral surveillance systems measure behaviors of interest in a large number of persons, but do not collect the same depth of information on specific topics that might be collected in smaller research studies.

Although many factors have been consistently reported to be associated with poor adherence (reviewed by Reynolds [9] and Ammasarri [10]), other associations are less clear. For many key demographic factors, as well as certain social and regimen-related factors, contradictory results appear in published literature [10]. Most recently, there have been contradictory reports as to whether some types of ARVT regimens (e.g., non-nucleoside reverse transcriptase inhibitor [NNRTI] -based regimens) may be associated with higher levels of adherence [11][16].

Methods for studying adherence are varied. They include electronic monitoring of access to medication bottles, examination of pharmacy records, physician estimation, and measurement of drug levels [17][21]. These methods are costly, however, and may be difficult to implement in large behavioral surveillance projects. Although self-reported adherence is subject to limitations common to interview studies and likely overestimates adherence [18], self-report provides a measure of adherence which generally correlates well with more objective methods [21][26] and with virologic treatment response [27]. It is also practical for use in ongoing behavioral surveillance efforts. We report the results of an exploratory analysis of behavioral surveillance data on self-reported adherence from a multi-state supplemental surveillance study of persons with HIV infection.



The Supplement to HIV and AIDS Surveillance (SHAS) Project is a cross-sectional behavioral surveillance study of persons with HIV infection. The methods have been previously described [28]. Briefly, adults, aged 18 years and older, reported with HIV or AIDS through routine case surveillance were eligible for participation. Participants were enrolled using one of two methods: 1) facility-based recruitment of all eligible persons seeking treatment at selected health care facilities in Denver, Colorado; Hartford and New Haven, Connecticut; Jacksonville, Miami, and Tampa, Florida; Atlanta, Georgia; Chicago, Illinois; Baltimore, Maryland; Detroit, Michigan; Jersey City and Paterson, New Jersey; and Philadelphia, Pennsylvania; or 2) population-based recruitment of all eligible persons in Phoenix and Tucson, Arizona; Delaware; Kansas; Los Angeles County, California; Minneapolis/St. Paul, Minnesota; New Mexico; South Carolina; Austin and Houston, Texas; and Washington. All sites interviewed persons reported with AIDS. Additionally, during the time period examined for this analysis, Arizona, Denver, Detroit, Florida, Kansas, Minnesota, New Jersey, New Mexico, South Carolina and Texas also interviewed persons with HIV infection which had not progressed to AIDS. Informed consent was obtained from all participants prior to the interview, and the study received institutional review board (IRB) approval at both the Centers for Disease Control and Prevention and local levels.


Beginning in May 2000, the SHAS Project added questions on medication history, adherence, and reasons for nonadherence. We utilized data from interviews conducted from May 2000 through June 2004 (when the project ended). Respondents who were prescribed antiretroviral therapy at the time of the interview were asked, “In the last 48 hours, have you missed or skipped any pills or spoonfuls of any of your antiretroviral medications?” We collected detailed information about the prescribed antiretroviral regimens at the time of the interview, including the number of daily doses prescribed for each medication. For respondents who reported missing one or more doses of medication in the 48 hours before interview, we asked about the number of doses missed, and calculated the total proportion of doses taken. We defined nonadherence as taking less than 95% of prescribed doses in the 48 hours before interview. We also asked the reasons why doses were missed and field coded the participants' responses using a list of possible reasons constructed by the investigators. Interviewers received training on administering the questionnaire, and used reference materials, including pictures of capsules of various medications, to help respondents identify their current prescriptions.

We defined protease inhibitor (PI)-based regimens as consisting of any 2 or more nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) plus any PI; boosted PI-based regimens as additionally including ritonavir; unboosted PI-based regimens as not including ritonavir; NNRTI-based regimens as consisting of any 2 or more NRTIs plus any NNRTI; triple NRTI therapy as consisting of any 3 or more NRTIs; and other ARVT as consisting of any one or more PI, NNRTI, NRTI, or pentafuside, but not meeting one of the previous definitions.

Analyses of All SHAS Respondents

We constructed a multivariate logistic regression model (PROC LOGISTIC, SAS Institute, Cary, NC) to calculate adjusted odd ratios and 95% confidence intervals for factors associated with nonadherence. We selected putative explanatory variables based on review of published biomedical literature. Our putative explanatory variables fell into several categories: demographic factors (age, race, sex, risk for acquiring HIV infection), socioeconomic status indicators (living situation, income, education, being on public assistance, and currently having health insurance) recent substance use behaviors (use of alcohol, crack, or injection drugs in the 12 months before the interview, and the CAGE questions [29] for possible alcohol abuse), characteristics of the ARVT regimen (type of regimen, number of drugs in the regimen, and duration of ARVT treatment), aspects of HIV care (use of a reminder system for medications, and whether the healthcare provider had discussed antiretroviral resistance with the patient), and indicators of depression (“feeling blue” for ≥14 of the past 30 days, having limited activity for ≥14 of the past 30 days, reporting a current met or unmet need for mental health services, and history of incarceration).

To construct our model, we first examined correlation coefficients of all pairs of putative explanatory variables. Where significant correlations existed, we removed one of the two correlated variables from consideration. For example, being on public assistance was removed from consideration, but current income was retained for consideration. We conducted bivariate analyses for all remaining explanatory variables, and only those with a p value of <0.25 from chi-square testing were considered in subsequent model building. Next we built a multivariate logistic regression model using manual forward selection. At each step, the remaining variable with the largest chi-square for its association with nonadherence was entered into the model, and inspected to determine if the associated p value in the model was <0.05. Factors continued to be entered until the last factor entered had a p value of >0.05. Once factors were entered into the model, they were not removed. The model was complete with respect to first order terms when no other variable could be entered and retained with a p value of <0.05. After all first order terms were entered into the model, we checked all two-way interaction terms, using an experiment-wide alpha of 0.05 because of the large number of two-way terms tested. No significant two-way interaction term was identified.

We conducted a similar process of modeling, in which, for certain factors with multiple levels – race, age, risk for HIV acquisition, ARVT regimen type, regimen duration, and number of drugs in regimen – if any level of the factor was significant, all levels of the factor would be retained in the model, regardless of significance. The results of this model were essentially the same as the reduced model, so we chose to present the reduced model.

For those who reported nonadherence, we also calculated the proportion of doses missed in the 48 hours before the interview, and tabulated the main reasons reported for missing doses.

Subgroup Analysis: Validation of Self-Reported Adherence Using Viral Load Measurements

We validated our self-reported adherence outcome measure by using data from a subset of SHAS respondents prescribed ARVT at the time of interview who were also observed in the Adult and Adolescent Spectrum of HIV Disease (ASD) Project (a longitudinal medical records review project conducted in some of the same project areas as SHAS) [30]. For those who were in both SHAS and ASD datasets, we determined the lowest HIV RNA concentration during the 6 month period including the SHAS interview date for each respondent. The distributions of HIV RNA concentration were not normal for either those who reported nonadherence or those who did not report nonadherence. Therefore, the median HIV RNA concentrations between the two groups were compared using the Wilcoxon Rank Sum test.


From May 2000 through June 2004, 11,503 persons were offered enrollment in SHAS; 9,088 (79%) were interviewed, and 2,415 (21%) refused to participate. Thirty-nine of 5,926 respondents who were prescribed ARVT at the time of interview had insufficient information on daily doses prescribed or number of doses missed in the 48 hours before the interview, and were, therefore, excluded. Thus, for this analysis, we included 5,887 (65% of interviewed) respondents who had complete information on antiretroviral prescription and dosing and who reported being prescribed at least one antiretroviral medication at the time of interview. Out of 5,887 respondents, 520 had been interviewed in a state where ASD was also conducted, and had data from medical records abstractions that spanned the date of the SHAS interview. The 520 were included in the subgroup analyses for validation of self-reported adherence using viral loads. The characteristics of interviewed persons, persons who were prescribed ARVT at the time of interview and included in the analysis of nonadherence, and persons included in the subgroup viral load analysis, are reported in Table 1.

Table 1. Characteristics of persons with HIV infection interviewed in the Supplement to HIV and AIDS Surveillance Project, who were prescribed antiretroviral drugs at the time of interview, and who were included in subgroup analyses of viral load, 18 US states, 2000–2004

Analyses of All SHAS Respondents

Among all SHAS respondents prescribed at least one antiretroviral medication at the time of the interview, 958 (16%) reported nonadherence in the 48 hours before the interview. The results of multivariate logistic regression (Table 2) indicated that several groups were significantly (p<0.05) more likely to report nonadherence: black non-Hispanic and Hispanic respondents (versus all other races); respondents aged 18–29 or 30–39 years (versus those aged ≥40 years); respondents who reported using alcohol or crack cocaine in the past 12 months; respondents prescribed ≥4 medications at the time of interview (versus 1–3 medications); respondents who had been blue for ≥14 of the past 30 days; and respondents currently living in a shelter or on the street (versus all other living situations). Respondents living in a medical facility were less likely to report nonadherence. Several additional factors were significantly associated with nonadherence, but had a weak or modest strength of association and lower bounds of the 95% CI close to 1.0. Such weak associations were observed for having both male-male sex and injecting drug use risks for HIV acquisition (versus all other acquisition modes); being prescribed an unboosted PI-based regimen (versus any other regimen); and being prescribed ARVT for ≥21 months (versus <21 months).

Table 2. Characteristics of HIV infected persons interviewed about adherence to currently prescribed antiretroviral therapy, number and proportion of nonadherent respondents, and logistic regression model of factors associated with nonadherence among 5,887 respondents to the Supplement to HIV/AIDS Surveillance Project, 18 US States, 2000 to 2004

Among the 958 persons who were nonadherent, the median proportion of doses missed was 50% of doses. The mode of missed doses was 100%; 226 of 958 (24%) of nonadherent respondents reported taking no dose of medication during the previous 48 hours.

A total of 821 persons (86% of those who missed ≥5% of doses in the 48 hours before the interview) reported the most important reason that doses were missed. The most commonly reported reasons were forgetting to take the medicine (266 respondents, 32%); side effects from the medications (152 respondents, 16%); inability to get to the clinic or doctor for a prescription (106 respondents, 11%); and inability to fit medications into the respondent's schedule (84 respondents, 9%). Other reported main reasons included, getting depressed (28 respondents, 3%); not believing that the medications work (21 respondents, 2%); and being on the street, in jail, or in prison (7 respondents, 1%).

Subgroup Analysis: Validation of Self-Reported Adherence Using Viral Load Measurements

Results from the subgroup who had HIV RNA concentrations available for analysis indicated that the HIV RNA concentration was higher for the 64 persons who reported taking <95% of prescribed doses in the 48 hours before the interview (median  =  400; range: undetectable – 425,000) than for the 397 persons who took ≥95% of prescribed doses (median  =  85; range: undetectable – 750,000; p  =  0.03 by 2-sided Wilcoxon Rank Sum Test). The HIV RNA measurements were taken a median of 36 days from the interview (interquartile range, 6–78 days).


In a large, diverse group of persons living with HIV infection, we observed a magnitude of nonadherence that poses a considerable threat to the individual and public health benefits of ARVT. About 1 in 6 persons interviewed in our project had missed at least 5% of their prescribed ARVT doses in the 48 hours before the interview. Although we used a self-reported measure, the self-reported measure of adherence was related to virologic outcome among a subgroup of our respondents, as indicated by viral load measurement – a relevant biological marker associated with adherence status [27], [31]. This suggests that self-report of recent adherence was correlated with prior adherence in our respondents.

Our findings provide additional data in some areas where inconsistent or null findings have been reported in the literature. For example, a summary analysis of published studies on adherence [10] reported that age and race had been inconsistently associated with adherence in previously published multivariate analyses. We report associations between adherence and both age and race; our data agree with another recent report of higher adherence among older persons living with HIV infection [32]. The same summary of published literature [10] identified several factors which had not generally been associated with adherence in the published literature. These included risk factor for HIV infection, length of time on therapy, and number of antiretroviral drugs. Our analysis showed statistically significant associations with each of these factors, although for risk factor and time on therapy, our associations were marginally significant, especially in light of our large sample size.

Many investigators have reported lower adherence among persons with current substance use [33][37]. In contrast, Martini et al. [38] found that heroin, cocaine, and alcohol users were not more likely to report nonadherence. In our data, alcohol use and crack use, but not recent injection drug use, were associated with nonadherence.

Depression is a factor which has also been consistently associated with nonadherence [2], [16], [37], [39][43]. Although we did not have data on clinical diagnoses of depression, we included putative explanatory variables which might be associated with depression. For example, days of “feeling blue”, self-identified need for mental health services, and days of limited activity were considered in our modeling. Despite consistent evidence of the association between adherence and depression in the literature and a significant association between nonadherence and higher number of days feeling blue in our model, only 3% of our nonadherent respondents identified getting depressed as a main reason for missing doses.

Among respondents in our project, prescription of an unboosted – but not a boosted – PI-based regimen, was weakly associated with nonadherence. Weiser et al. [13] reported no difference in adherence between patients prescribed NNRTI versus PI-based regimens. Other researchers have reported lower adherence among patients prescribed PI-based regimens [12], [14], [16], while Glass et al. reported more nonadherence among patients prescribed a boosted PI-based regimen [15]. If adherence is truly higher in patients prescribed NNRTI-based regimens, it may be explained by a lower pill burden [14], [33], [44]. However, our analysis partially controlled for lower pill burden by controlling for the number of antiretroviral medications prescribed. Alternatively, higher adherence for those prescribed NNRTI regimens could be explained by fewer intolerable side effects. In our data, side effects were a commonly reported reason for missing doses among those who were nonadherent. Further, self-perception of body changes such as peripheral fat loss or central fat gain are associated with lower adherence [45], and may be more common among those taking PI-based ARVT [46].

Even if our finding of a marginal association of nonadherence with prescription of an unboosted PI-based regimen were confirmed by others, the clinical significance of this finding would be unclear. Recent findings [47] suggest that viral suppression may be maintained with less than 95% adherence for patients on boosted PI-based regimens. If this finding is substantiated in other settings, then future analyses of adherence and clinical guidelines may consider having regimen-specific thresholds to define acceptable adherence [48].

Pill burdens for patients currently on ARVT are likely to be significantly lower now than were pill burdens at the time of our interviews [33], [44]. This is because of new formulations which include multiple drugs in a single pill or capsule, and regimens which require fewer prescribed doses per day.

Although the most commonly reported reason for missed ARVT doses was forgetting to take medications, the modal proportion of doses missed was 100%. The fact that nearly a quarter of non-adherent respondents took no medication during the 48 hour period suggests that, in many cases, doses are not missed sporadically. More prolonged periods of missed doses could represent “drug holiday” periods [49], periods in which patients ran out of medication, or periods when significant side effects or illness prevented taking any. Future work should attempt to characterize the reasons for these prolonged periods of missed doses.

Our study has some important limitations. As with any interview study, our data are subject to recall bias and social desirability bias [50]. We believe asking about adherence in the 48 hours before the interview should minimize recall bias. Social desirability bias should lead to under-report of poor adherence, which has been recently confirmed [18]; this suggests that our data represent minimum estimates of nonadherence. However, social desirability bias has been reported not to be a major source of error in collecting self-reported adherence data [51]. The persons recruited and interviewed in the SHAS project are not representative of all persons with HIV infection or AIDS in the project areas, or in the United States.

Further, there are important factors which are likely related to nonadherence which the SHAS interview did not collect, and therefore could not be included in our analysis. For example, we did not collect information on social support or isolation [52] (other than current living arrangements [39]), beliefs about prognosis or treatment efficacy [53], [54], perceptions of body changes [45], cognitive factors [55], [56], serostatus disclosure [57], or life stress [54], [58].

These results have some relevance for clinicians and public health officials. Based on our findings and existing treatment guidelines [59], providers should consider that certain groups of patients may be especially likely to benefit from support and guidance around the importance of adherence and strategies for improving adherence. Our data, and the data of others, support referral to housing resources for homeless patients and referral to appropriate treatment for alcohol abuse, crack cocaine use, and mental health problems, either before or concurrently with prescription of antiretroviral medications. Some other factors associated with nonadherence among our respondents, such as race/ethnicity and age, are likely markers for other barriers that lead to poor adherence. Additional focused research is needed to determine the underlying challenges to adherence for these patients, and to develop interventions and resources to help physicians and patients overcome these challenges.

Adherence to antiretroviral medications also has important public health implications. Poor adherence to ARVT may be associated with the development of drug-resistant strains of HIV. Also, to the extent that ARVT lowers prevalent viral loads on a population basis, there should be a desirable impact on reducing the risk of HIV transmission. Poor adherence also compromises this potential population-level benefit of ARVT. Thus, adherence is a relevant behavior for targeting with interventions, and for monitoring as a part of future behavioral surveillance projects [60]. Although large-scale behavioral surveillance efforts are not designed to provide detailed clinical information about adherence and its consequences, and are subject to some important limitations, such systems offer the opportunity to measure important variables related to HIV care, including adherence, in an ongoing and systematic way.


The authors thank Eric Schwab, MS for his technical and editorial assistance, and Drs. Amy Lansky and Travis Sanchez for critical review. We thank April Smith for editorial assistance.

We also gratefully acknowledge the contributions of the SHAS Project principal investigators and project officers: Vjollca Berisha and Rick DeStephens, Arizona Department of Public Health; Arthur Davidson, Denver Health and Hospital Authority; Aaron Roome, Connecticut Department of Public Health; James Welch, Delaware Department of Public Health; Rebecca Grigg and Pam Lowell, Florida Department of Health; Jeffrey Lenox and Alan Fann, Emory University School of Medicine; Marcia Wolverton, Houston Department of Health and Human Services; Fran Eury, Illinois Department of Public Health; Jeni Mulqueen, Anthony Merriweather, and Gail Hansen, Kansas Department of Health and Environment; Amy Rock Wohl and Denise Johnston, Los Angeles County Department of Health; Ellen Caldeira, Maryland Department of Health and Mental Hygiene; Eve Mokotoff, Sha Juan Colbert, and Shanell McGoy, Michigan Department of Community Health; Richard Danila and Don Stiepan, Minnesota Department of Public Health; Sally D'Errico, New Jersey Department of Health; Mack Sewell, New Mexico Department of Public Health; Kathleen Brady, Philadelphia Department of Public Health; Terri Stephens, South Carolina Department of Health and Environmental Control; Sharon Melville and Sylvia Odem, Texas Department of Health; and Maria Courogen, Washington State Department of Health.

Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.

Author Contributions

Conceived and designed the experiments: PS MC EB AN. Performed the experiments: PS MC EB AN. Analyzed the data: PS GN JS. Contributed reagents/materials/analysis tools: MC GN EB JS AN. Wrote the paper: PS.


  1. 1. Carpenter CC, Fischl MA, Hammer SM, Hirsch MS, Jacobsen DM, et al. (1998) Antiretroviral therapy for HIV infection in 1998: updated recommendations of the International AIDS Society-USA Panel. JAMA 280: 78–86.
  2. 2. Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, et al. (2000) Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med 133: 21–30.
  3. 3. Hirsch MS, Conway B, D'Aquila RT, Johnson VA, Brun-Vezinet F, et al. (1998) Antiretroviral drug resistance testing in adults with HIV infection: implications for clinical management. International AIDS Society–USA Panel. JAMA 279: 1984–1991.
  4. 4. Wood E, Hogg RS, Yip B, Harrigan PR, O'Shaughnessy MV, et al. (2004) The impact of adherence on CD4 cell count responses among HIV-infected patients. J Acquir Immune Defic Syndr 35: 261–268.
  5. 5. Cohen M Antiretroviral therapy to prevent sexual transmission of HIV. 13th Conference on Retroviruses and Opportunistic Infections, Denver Colorado, February 2006. Abstract 54.
  6. 6. Weiser B, Nachman S, Tropper P, Viscosi KH, Grimson R (1994) Quantitation of human immunodeficiency virus type 1 during pregnancy: relationship of viral titer to mother-to-child transmission and stability of viral load. Proc Natl Acad Sci USA 91: 8037–8041.
  7. 7. Vlahov D, Safaien M, Lai S, Strathdee SA, Johnson L, et al. (2001) Sexual and drug risk-related behaviours after initiating highly active antiretroviral therapy among injection drug users. AIDS 15: 2311–2316.
  8. 8. Lansky A, Sullivan PS, Gallagher K, Fleming PL (2007) The HIV behavioral surveillance program in the US: a conceptual framework. Public Health Reports 122: Supp116–23.
  9. 9. Reynolds NR (2004) Adherence to antiretroviral therapies: state of the science. Curr HIV Res 2: 207–214.
  10. 10. Ammassari A, Trotta MP, Murri R, Castelli F, Narciso P, et al. (2002) Correlates and predictors of adherence to highly active antiretroviral therapy: overview of published literature. J Acquir Immune Defic Syndr 31: Suppl 3S123–S127.
  11. 11. Trotta MP, Ammassari A, Cozzi-Lepri A, Zaccarelli M, Castelli F, et al. (2003) Adherence to highly active antiretroviral therapy is better in patients receiving non-nucleoside reverse transcriptase inhibitor-containing regimens than in those receiving protease inhibitor-containing regimens. AIDS 17: 1099–1102.
  12. 12. Mannheimer S, Friedland G, Matts J, Child C, Chesney M (2002) The consistency of adherence to antiretroviral therapy predicts biologic outcomes for human immunodeficiency virus-infected persons in clinical trials. Clin Infect Dis 34: 1115–1121.
  13. 13. Weiser SD, Guzman D, Riley ED, Clark R, Bangsberg DR (2004) Higher rates of viral suppression with nonnucleoside reverse transcriptase inhibitors compared to single protease inhibitors are not explained by better adherence. HIV Clin Trials 5: 278–287.
  14. 14. Maggiolo F, Ravasio L, Ripamonti D, Gregis G, Quinzan G, et al. (2005) Similar adherence rates favor different virologic outcomes for patients treated with nonnucleoside analogues or protease inhibitors. Clin Infect Dis 40: 158–163.
  15. 15. Glass TR, De GS, Weber R, Vernazza PL, Rickenbach M, et al. (2006) Correlates of self-reported nonadherence to antiretroviral therapy in HIV-infected patients: the Swiss HIV Cohort Study. J Acquir Immune Defic Syndr 41: 385–392.
  16. 16. Carrieri MP, Leport C, Protopopescu C, Cassuto JP, Bouvet E, et al. (2006) Factors associated with nonadherence to highly active antiretroviral therapy: a 5-year follow-up analysis with correction for the bias induced by missing data in the treatment maintenance phase. J Acquir Immune Defic Syndr 41: 477–485.
  17. 17. Bartlett JA (2002) Addressing the challenges of adherence. J Acquir Immune Defic Syndr 29: Suppl 1S2–10.
  18. 18. Murri R, Ammassari A, Trotta MP, De LA, Melzi S, et al. (2004) Patient-reported and physician-estimated adherence to HAART: social and clinic center-related factors are associated with discordance. J Gen Intern Med 19: 1104–1110.
  19. 19. Fairley CK, Permana A, Read TR (2005) Long-term utility of measuring adherence by self-report compared with pharmacy record in a routine clinic setting. HIV Med 6: 366–369.
  20. 20. Walsh JC (1999) Adherence in clinical trials and in clinical practice. Antivir Ther 4: Suppl 349–54.
  21. 21. Kerr T, Walsh J, Lloyd-Smith E, Wood E (2005) Measuring adherence to highly active antiretroviral therapy: implications for research and practice. Curr HIV/AIDS Rep 2: 200–205.
  22. 22. Chesney MA, Ickovics JR, Chambers DB, Gifford AL, Neidig J, 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: 255–266.
  23. 23. Fletcher CV, Testa MA, Brundage RC, Chesney MA, Haubrich R, et al. (2005) Four measures of antiretroviral medication adherence and virologic response in AIDS Clinical Trials Group Study 359. J Acquir Immune Defic Syndr 40: 301–306.
  24. 24. Antinori A, Cozzi-Lepri A, Ammassari A, Trotta MP, Nauwelaers D, et al. (2004) Relative prognostic value of self-reported adherence and plasma NNRTI/PI concentrations to predict virological rebound in patients initially responding to HAART. Antivir Ther 9: 291–296.
  25. 25. Walsh JC, Mandalia S, Gazzard BG (2002) Responses to a 1 month self-report on adherence to antiretroviral therapy are consistent with electronic data and virological treatment outcome. AIDS 16: 269–277.
  26. 26. Simoni JM, Hurth A, Pearson C, Panatalone D, Merril J, et al. (2006) Self-reported measures of antiretroviral therapy adherence: a review with recommendations for HIV research and clinical management. AIDS Behav 10: 199–212.
  27. 27. Nieuwkerk PT, Oort FJ (2005) Self-reported adherence to antiretroviral therapy for HIV-1 infection and virologic treatment response: a meta-analysis. J Acquir Immune Defic Syndr 38: 445–448.
  28. 28. Buehler JW, Diaz T, Hersh BS, Chu SY (1996) The supplement to HIV-AIDS Surveillance project: an approach for monitoring HIV risk behaviors. Public Health Rep 111: Suppl 1133–7.133–137.
  29. 29. Maisto SA, Saitz R (2003) Alcohol use disorders: screening and diagnosis. Am J Addict 12: Suppl 1S12–S25.
  30. 30. Farizo KM, Buehler JW, Chamberland ME, Whyte BM, Froelicher ES, et al. (1992) Spectrum of disease in persons with human immunodeficiency virus infection in the United States. JAMA 267: 1798–1805.
  31. 31. Vrijens B, Goetghebeur E, de Klerk E, Rode R, Mayer S, et al. (2005) Modelling the association between adherence and viral load in HIV-infected patients. Stat Med 24: 2719–2731.
  32. 32. Silverberg MJ, Leyden W, Horberg MA, DeLorenze GN, Klein D, et al. (2007) Older age and the response to and tolerability of antiretroviral therapy. Arch Intern Med 167: 684–691.
  33. 33. Lafeuillade A (2001) Factors affecting adherence and convenience in antiretroviral therapy. Int J STD AIDS 12: Suppl 418–24.
  34. 34. Berg KM, Demas PA, Howard AA, Schoenbaum EE, Gourevitch MN, et al. (2004) Gender differences in factors associated with adherence to antiretroviral therapy. J Gen Intern Med 19: 1111–1117.
  35. 35. Moss AR, Hahn JA, Perry S, Charlebois ED, Guzman D (2004) Adherence to highly active antiretroviral therapy in the homeless population in San Francisco: a prospective study. Clin Infect Dis 39: 1190–1198.
  36. 36. Howard AA, Arnsten JH, Lo Y, Vlahov D, Rich JD, et al. (2002) A prospective study of adherence and viral load in a large multi-center cohort of HIV-infected women. AIDS 16: 2175–2182.
  37. 37. Arnsten JH, Demas PA, Grant RW, Gourevitch MN, Farzadegan H, et al. (2002) Impact of active drug use on antiretroviral therapy adherence and viral suppression in HIV-infected drug users. J Gen Intern Med 17: 377–381.
  38. 38. Martini M, Recchia E, Nasta P, Castanotto D, Chiaffarino F, et al. (2004) Illicit drug use: can it predict adherence to antiretroviral therapy? Eur J Epidemiol 19: 585–587.
  39. 39. Gordillo V, del Amo J, Soriano V, Gonzalez-Lahoz J (1999) Sociodemographic and psychological variables influencing adherence to antiretroviral therapy. AIDS 13: 1763–1769.
  40. 40. Ammassari A, Antinori A, Aloisi MS, Trotta MP, Murri R, et al. (2004) Depressive symptoms, neurocognitive impairment, and adherence to highly active antiretroviral therapy among HIV-infected persons. Psychosomatics 45: 394–402.
  41. 41. Starace F, Ammassari A, Trotta MP, Murri R, De LP, et al. (2002) Depression is a risk factor for suboptimal adherence to highly active antiretroviral therapy. J Acquir Immune Defic Syndr 31: Suppl 3S136–S139.
  42. 42. Sledjeski EM, Delahanty DL, Bogart LM (2005) Incidence and impact of posttraumatic stress disorder and comorbid depression on adherence to HAART and CD4+ counts in people living with HIV. AIDS Patient Care STDS 19: 728–736.
  43. 43. Chander G, Himelhoch S, Moore RD (2006) Substance abuse and psychiatric disorders in HIV-positive patients : epidemiology and impact on antiretroviral therapy. Drugs 66: 769–789.
  44. 44. Piliero PJ, Colagreco JP (2003) Simplified regimens for treating HIV infection and AIDS. J Am Acad Nurse Pract 15: 305–312.
  45. 45. Santos CP, Felipe YX, Braga PE, Ramos D, Lima RO, et al. (2005) Self-perception of body changes in persons living with HIV/AIDS: prevalence and associated factors. AIDS 19: Suppl 4S14–S21.
  46. 46. Lichtenstein KA (2005) Redefining lipodystrophy syndrome: risks and impact on clinical decision making. J Acquir Immune Defic Syndr 39: 395–400.
  47. 47. Gross R, Yip B, Wood E, Bangsberg D, Justice A, Montaner J, et al. Boosted PIs are more forgiving of suboptimal adherence than non-boosted PIs or NNRTIs. Thirteenth Conference on Retroviruses and Opportunistic Infections, Denver, abstract 533, 2006.
  48. 48. Smith RJ (2006) Adherence to antiretroviral HIV drugs: how many doses can you miss before resistance emerges? Proc Biol Sci 273: 617–624.
  49. 49. Van Wijngaerden E, De Saar V, DeGraeve V, Vandamme AM, Van Vaerenbergh K, et al. (2002) Nonadherence to highly active antiretroviral therapy: clinically relevant patient categorization based on electronic event monitoring. AIDS Res Hum Retroviruses 18: 327–330.
  50. 50. Sackett DL (1979) Bias in analytic research. J Chronic Dis 32: 51–63.
  51. 51. Wagner G, Miller LG (2004) Is the influence of social desirability on patients' self-reported adherence overrated? J Acquir Immune Defic Syndr 35: 203–204.
  52. 52. Altice FL, Mostashari F, Friedland GH (2001) Trust and the acceptance of and adherence to antiretroviral therapy. J Acquir Immune Defic Syndr 28: 47–58.
  53. 53. Holmes WC, Pace JL (2002) HIV-seropositive individuals' optimistic beliefs about prognosis and relation to medication and safe sex adherence. J Gen Intern Med 17: 677–683.
  54. 54. Nilsson SL, Diamond PM, Ross MW, Williams M, Bratt G (2006) Baseline predictors of three types of antiretroviral therapy (ART) adherence: a 2-year follow-up. AIDS Care 18: 246–253.
  55. 55. Wagner GJ (2002) Predictors of antiretroviral adherence as measured by self-report, electronic monitoring, and medication diaries. AIDS Patient Care STDS 16: 599–608.
  56. 56. Selnes OA (2002) Neurocognitive aspects of medication adherence in HIV infection. J Acquir Immune Defic Syndr 31: Suppl 3S132–S135.
  57. 57. Stirratt MJ, Remien RH, Smith A, Copeland OQ, Dolezal C, et al. (2006) The role of HIV serostatus disclosure in antiretroviral medication adherence. AIDS Behav 10: 483–493.
  58. 58. Bottonari KA, Roberts JE, Ciesla JA, Hewitt RG (2005) Life stress and adherence to antiretroviral therapy among HIV-positive individuals: a preliminary investigation. AIDS Patient Care STDS 19: 719–727.
  59. 59. Department of Health and Humans Services (2006) Guidelines for the use of antiretroviral agents in HIV-1 infected adults and adolescents: adherence to potent antiretroviral therapy. Available at: Last accessed: May 31, 2007.
  60. 60. Gallagher KM, Sullivan PS, Lansky A, Onorato IM (2007) Behavioral surveillance among persons risk for HIV infection in the United States: The National HIV Behavioral Surveillance System. Public Health Reports 122: Supp 132–38.