The introduction of protease inhibitor-based combination therapies has had a major impact on the quality of life and the prospects for an extended survival in HIV-infected persons; many of them have been able to go back to work and lead a full social life[1,2]. Although providing these benefits, combination therapy can also make extraordinary demands on an individual due to their often complicated regimens and side effects. Key to the success of the new highly active antiretroviral therapies (HAART) are the ability and willingness of HIV-positive individuals to adhere to complex antiretroviral regimens. The consequences of poor adherence include not only diminished benefits for the patient, but also the public health threat of the emergence of multidrug-resistant viruses, as these resistant strains can then be transmitted from a patient to their contacts[4-6]. The new generation of HIV drugs offers the potential for long-term suppression of HIV replication; however, the challenge now is to encourage and enable patients to take these medicines correctly, in order to achieve their maximum effect.
Evaluating adherence has proven to be difficult and most studies have been based on pill counts[3,8,9]. On the other hand, when examining the predictors of compliance, some authors have recommended taking into account psychological factors such as stress, depression and coping skills since they have been shown to decrease adherence to the prescribed medication[10-12]. Few studies have emphasized the patients‚ perception of their illness and medication, the availability of social support, patients‚ understanding of the rationale for treatment and the patient-doctor relationship[7,13]. Both sociological and psychological factors as determinants of adherence have been seldom considered, but if adherence is related to the ability and willingness of HIV-positive individuals to take complex antiretroviral regimens correctly, there is no doubt that these variables should play an important role.
We have examined which and to what extend sociodemographic and psychological factors influence the adherence to antiretroviral therapy in HIV-infected patients treated in a single HIV/AIDS institution, in order to develop strategies to improve compliance.
Patients and methods
All HIV-infected individuals treated with antiretroviral drugs and attending an HIV/AIDS reference institution located in Madrid during a 6-month period (December 1997 to May 1998) were considered eligible for this study.
All patients were asked for voluntary completion of a questionnaire which recorded information on different sociodemographic and psychological variables as well as data on drug compliance. Clinical charts were inspected in order to collect additional clinical information from the patients.
The questionnaire was developed by the research team, using items available from questionnaires such as the Beck Depression Inventory (BDI) and Hamilton Anxiety score test  after discussion with a panel of experts. The questionnaire included 47 questions plus the Beck Inventory Questionnaire. There were three questions addressing the patient-doctor relationship, four regarding knowledge and beliefs about the illness and 10 regarding social support. The questionnaire was piloted on 20 individuals and modified accordingly. The definitive questionnaire recorded data on sociodemographic characteristics, social support, patient- doctor relationship, knowledge about the illness, beliefs about treatment and the perceived health situation. We also introduced questions regarding the understanding of the prescribed regimen and difficulties in taking the different drugs as well as their causes. The last part of the questionnaire was a daily schedule in which the patient had to indicate when and how he/she took each of the drugs included in the medication regimen. The questionnaire, filled out with the help of a trained psychologist, was anonymous and patients were informed that its contents would not be passed on to their physicians.
Adherence to treatment was assessed using patient‚s self-report and the pill count method. Instead of giving as index of good adherence values above 80%, which is a threshold based on data from other chronic illnesses such as hypertension or diabetes, it was decided to define satisfactory or good adherence when >90% of the pills prescribed in any regimen were taken in accordance with the prescribed regime, in agreement with the suggestions from recent HIV literature[12,17-19], and following the recommendations made by the International AIDS Society-USA Panel Guidelines, in which it is stated that ‚less than excellent adherence may result in virus breakthrough and the emergence of drug-resistant strains‚. The adherence assessment was performed according to the coincidence of the prescribed regimen to the daily schedule filled out by the patient (how many pills and when they were taken during the past week), and the pill count method in a subgroup of 115 patients, who were asked to return those pills that had not been taken in the previous month to the pharmacist.
Depression was examined using the Beck Depression Inventory (BDI), which has been successfully used to assess the depression level in HIV-infected patients in previous studies[10,21-24]. Although three categories are classically distinguished in the depression scale: slight (10-15 points), moderate (16-23), and severe (24-62), a cut-off of 14 was used in our study to distinguish between depressed and non-depressed individuals 
The level of anxiety was measured through questions relating to the fear of being hospitalized, focusing on the ‚state of anxiety‚ more than on the ‚trait of anxiety‚. Questions were developed by modifying the Hamilton Anxiety score test. The L variable was a variable constructed out of 5 items to asses the internal consistency of the answers given by patients. When more than 80% of discrepancies were found in their responses, it was felt that patients did not want to disclose their situation, including their adherence to treatment, or were unable to do so. Analyses were carried out both including and excluding those patients and there were no differences in the results. All patients are included in the final analyses.
Clinical data, CD4 lymphocyte count, plasma viral load, frequency of medication and the time period of continuing antiretroviral therapy were recorded by asking the patients and confronting this information with that recorded in the clinical chart. The CD4 cell count was measured using flow cytometry (Coulter, New Jersey, USA), and the viral load was quantified using the second generation branched DNA assay (Chiron, California, USA), which has a detection limit of 500 copies per ml.
Data were analysed using the SPSS version 7.5 (SPSS Inc. Chicago, Illinois, USA) and STATA (Stata Corp., San Antonio, Texas, USA). Comparisons were performed using the Student‚s t test for normally distributed data, the Mann-Whitney test for not normally distributed data and the χ2 test for categorical data. The magnitude of the association between the different variables in relation to the adherence to treatment was measured through odds ratios (OR) and their 95% confidence intervals (CI). Comparisons for which P-values were below 0.05 were considered significant. Mantel-Haenszel and multiple logistic regression analyses were performed to find the best predictive model and to test for effect modification. The best predictive model for compliance with antiretroviral therapy was built through a backward elimination procedure.
Out of a total of 371 eligible HIV-positive patients, five were excluded because they did not complete the questionnaire, leaving a sample of 366 subjects. The descriptive characteristics of the population are summarized in Table 1. Of these, 279 (76.2%) were men. The median age at enrolment into the study was 35 years old (range 20-72). The route of infection was intravenous drug use (IVDU) in 162 (44.3%) subjects but only 65 (18%) admitted to be current IVDU. Only 12 (3.3%) had no school education and 66 (18%) were unemployed at the time they were interviewed.
For most patients, 355 (97%), more than one antiretroviral drug had been prescribed at the time they were interviewed: 85 (23%) were on double, 252 (69%) on triple and 18 (5%) on quadruple combinations. The median time that they had been under antiretroviral treatment was 30 months. There was no statistically significant association between compliance and the length of time that patients had been on antiretroviral treatment for. Only 74 (21.4%) individuals had CD4 cell counts below 200×106 cells/l and 213 (58.2%) had undetectable levels of plasma HIV RNA.
Depression was found in 148 (40.4%) of patients and 94 (25.7%) perceived they had no social support. Significant discrepancies between answers (lying) were observed in 71 (19.4%) patients; however, anxiety was identified in 155 (42.4%) subjects. The level of knowledge about the illness and beliefs about therapy were judged to be satisfactory by 263 (71.8%) and 338 (92.3%) individuals, respectively.
Variables associated with compliance: univariate analysis
Compliance with antiretroviral treatment was considered satisfactory (>90% consumption of prescribed pills) in 211 (57.6%) patients.
Table 2 shows the crude OR for variables identified as predictors of compliance which were statistically significant in the univariate analysis. Unsatisfactory (or bad) compliance was commoner among younger individuals, whereas the best compliance was seen in subjects aged between 32 and 35 years. By considering the HIV exposure category, and given that homosexuals and heterosexuals had similar adherence rates, the population was classified into two groups, IVDUs and non-IVDUs. A better compliance was recorded for the last group (OR, 2.13; 95% CI, 1.38-3.22).
An association between compliance with therapy and level of studies was observed: subjects with a low level of education had the worst adherence, whereas individuals with a university degree had the best (see Fig. 1). Having a job was associated with better compliance (OR, 2.24; 95% CI, 1.27-2.73) as was the perception of a good social support (OR, 2.03; 95% CI, 1.26-3.27). Lack of depression was associated with better adherence (OR, 1.79; 95% CI, 1.27-2.73).
Level of CD4 cell count was associated with compliance; worse compliance was commoner among subjects with CD4 cell counts below 200cells×106/l and in those 20 subjects with unknown CD4 count values. Adherence was better in subjects with a CD4 count over 200 cells×106/l and these differences were statistically significant.
Other variables, which were referred to knowledge about the illness, beliefs about therapy and satisfaction with the clinician were not statistically associated with compliance to antiretroviral drugs. Viral load values and the length of time patients had been on treatment for were not statistically associated with adherence.
Variables associated with compliance: multivariate analysis
In the multivariate analysis, the best predictive model for adherence included the transmission category, age, CD4 cell count at enrolment, depression, social support and the interaction between the last two variables (P value for the interaction, 0.008) (see Table 3). IVDUs and individuals that were less than 32 years of age had the poorest compliance with the prescribed treatment. In contrast, the best adherence was recorded among subjects aged 32-35 (OR, 2.31; 95% CI, 1.21-4.40). Non-IVDUs had a compliance which was more than twice that of IVDUs (OR, 2.05; 95% CI, 1.28-3.29) and subjects with CD4 cell counts from 200-499×106 cells/l at enrolment had better compliance than those with CD4 cell counts below 200×106 cells/l (OR, 2.78; 95%CI, 1.40-5.51).
Finally, subjects without depression and good social support had a compliance which was nearly twice that of depressed subjects who lacked social support (OR, 1.86; 95% CI, 0.98-3.53) although the lower limits of the confidence intervals were just below one. Compliance was poorer in the remaining three groups; depressed individuals irrespectively of their social support and subjects who were not depressed but who lacked social support.
This study found an overall prevalence of satisfactory compliance with antiretroviral drugs of 57.6%, which is low but similar to that found in other studies, although some have used a different criteria to define good adherence[7,8,26,27]. We have also identified a number of sociodemographic and psychological variables which were associated with compliance in the univariate analysis such as age, transmission category, level of education, work situation, CD4 cell count level, depression and self-perceived social support. In the multivariate model, however, only age, transmission category, CD4 cell count level, depression, self-perceived social support, and an interaction between the last two variables predicted a satisfactory adherence to treatment.
The assessment of compliance in this study was done through patients‚ self-report and the pill count method in a sub-group of patients. Although self-reporting could overestimate the rate of satisfactory adherence to the medication[28,29], the results obtained in a smaller sample of the study population, using a supervised pill count method, gave a similar level of adherence to treatment. Some authors have suggested that self-report has the potential to be one of the most accurate measures of behavioural adherence because only the patient can report actual behaviour.
A good compliance was more often recorded in non-IVDUs, subjects aged between 32 and 35 years, those with CD4 cell count values over 200×106 cells/l and those who were not depressed and had adequate social support. Previous studies have suggested that unsatisfactory compliance is commoner among IVDUs [27,30,31] and younger individuals [12,32] but the importance of psychological variables such as depression and self-perceived social support has been less reported.
In the univariate analyses, both depression and lack of social support were associated with worse compliance with treatment but in the multivariate analysis, the interaction found between depression and perceived social support suggests that social support does not contribute further to improve adherence in the presence of depression. Depressed individuals comply worse with treatment irrespectively of the social support they have whereas in subjects who are not depressed, social support does contribute to improved adherence to treatment. This implies that efforts to provide social services to HIV-positive subjects should run in parallel with adequate management of depression by multidisciplinary teams. Other variables identified as predictors of compliance in the univariate analyses such as the level of education and the work situation were no longer predictors in the multivariate model.
The objective of this study was to identify sociodemographic and psychological variables that predict poor adherence to antiretroviral therapy. This knowledge should provide an opportunity for the design of new strategies to improve adherence to the new potent drug combinations. Depressive symptoms are common and often underdiagnosed among HIV-positive patients, and depression has been associated with increased reports of HIV-related symptoms[33,34]. Although depressive symptoms have not been associated with an accelerated HIV disease progression in some studies, others have found an association[34,36], and depression has been identified as an indicator of a shorter survival period in HIV-positive men[37,38]. This effect on survival seems to be independent from other immunological and clinical variables associated with disease progression and might be mediated through mechanisms such as poorer health habits and reduced adherence to treatment.
The design of the study did not allow us to establish a causal association between depression and poor compliance, as both variables were measured at the same time. However, although poor compliance hypothetically might lead to depression, the opposite is more convincing, since depressed individuals may be less motivated to take their medication. This hypothesis is currently being explored, but has been recently pointed out by others[39,40]. In any case, awareness of depression among health professionals caring for HIV-positive patients needs to be increased in order to intervene appropriately.
The association found between level of CD4 cell counts at enrolment and compliance to therapy suggests that individuals with higher CD4 cell counts tend to have better compliance than those with lower levels or unknown values but again, as this was a cross-sectional study, we cannot imply that individuals with more advanced disease have worse compliance because it could also mean that those who comply worse have a more protracted course of their HIV-infection. Again, we will be able to explore this further in follow-up studies of this cohort.
The improvement in the quality of patients‚ life and motivation for adhering to treatment is influenced by the behaviour of both patient and clinician although in our study, this was not associated with better adherence to treatment. They are strongly related, and could be enhanced with the collaboration of behavioural research. This fact challenges doctors to collaborate with professionals outside medicine to explore the gap between intention and action. If the objective of antiretroviral therapy is to prolong life, prevent disease progression and improve the quality of life, it is critical that patients are provided with as much education and support as possible, given the need for maximal long term adherence to currently available treatments. Beyond the obvious medical, pharmacokinetic, and safety issues, the patient‚s lifestyle, motivations and resources should also be considered.
Knowledge about the illness, belief in the treatment and satisfaction with the clinician were high in our sample but were not associated with compliance. Since this institution is a tertiary referral centre with no defined catchment area and open and free access, it is likely that there is a selection process by which better-informed patients attend this hospital. This is also suggested by the fact that only 44.3% of the patients in this study were IVDUs, whereas in Madrid 67% of AIDS patients are IVDUs. In addition, a high proportion of all the IVDUs in our study population were not current injectors which may affect the generalizability of these results to other centres with a different population.
Many of the new strategies for improving adherence to antiretroviral drugs seek to improve the patients‚ motivation, which can only be achieved by listening to the whole patient (‚other issues that affect the patient, affect adherence‚)[44,45]. The implementation of these strategies in the real world requires multidisciplinary efforts and counselling by trained psychologists may be necessary in this context since these professionals can provide guidance and be able to offer some assistance, treat depression, and ultimately improve adherence.
1. Palella F, Delaney K, Moorman A, et al
. Declining morbidity and mortality among patients with advanced HIV infection.N Engl J Med
2. Lange J. Issues for the future of antiretroviral therapy
. Antiviral Ther
3. Williams A, Friedland G. Adherence, compliance, and HAART.AIDS Clin Care
4. Veenstra J, Schuurman R, Cornelissen M, et al
. Transmission of zidovudine-resistant HIV-1 following deliberate injection of blood from a patient with AIDS: characteristics and natural history of the virus.Clin Infect Dis
5. Imrie A, Beveridge A, Genn W, Vizzard J, Cooper D. Transmission of HIV-1 resistant to nevirapine and zidovudine: Sydney Primary HIV Infection Study Group.J Infect Dis
6. Hecht F, Grant R, Petropoulos C, et al. Sexual transmission of an HIV-1 variant resistant to multiple reverse-transcriptase and protease inhibitors.N Engl J Med
7. Erwin J. Adherence and its implications for antiretroviral therapy.Int Antiviral News
8. Rodriguez-Rosado R, Jiménez-Nácher I, Soriano V, Antón P, González-Lahoz J. Virological failure and adherence to antiretroviral therapy in HIV-infected patients [letter]. AIDS
9. Piñeiro F, Gil V, Donis M, et al
. Usefulness of six indirect methods to evaluate drug therapy compliance in non-insulin-dependent diabetes mellitus.Rev Clin Esp
10. Singh N, Squier C, Sivek C, et al
. Determinants of compliance with antiretroviral therapy in patients with human immunodeficiency virus: prospective assessment with implications for enhancing compliance.AIDS Care
11. Rabkin J, Ferrando S. A “second life“ agenda. Psychiatric research issues raised by protease inhibitor treatments for people with the human immunodeficiency virus or the acquired immunodeficiency syndrome.Arch Gen Psychiatry
12. Mehta S, Moore RD, Graham NM. Potential factors affecting adherence with HIV therapy. AIDS
13. Besch C. Compliance in clinical trials. AIDS
14. Caraël M, Coates TJ. Social, cultural and political aspects.AIDS
15. Conde Y, Useros S. Adaptación castellana de la escala de evaluación conductual para la depresión de Beck.Rev Psicol y Psiq Méd
16. Hamilton M. The assessment of anxiety states by raring,Brit J Med Psychol
17. Jong MD, De Boer RJ, De Wolf F, et al
. Overshoot of HIV-1 viraemia after early discontinuation of antiretroviral treatment.AIDS
18. Fätkenheuer G, Theisen A, Rockstroh J, et al
. Virological treatment failure of protease inhibitor therapy in an unselected cohort of HIV-infected patients.AIDS
19. Montaner J, Reiss P, Cooper D, et al
. A randomized, double-blind trial comparing combinations of nevirapine, didanosine, and zidovudine for HIV-infected patients. JAMA
20. Carpenter C, Fischl M, Hammer S, et al
. Antiretroviral therapy for HIV infection in 1997.JAMA
21. Leserman J, Petitto JM, Perkins D, et al
. Severe stress, depressive symptoms, and changes in lymphocyte subsets in human immunodeficiency virus-infected men. A 2-year follow-up study.Arch Gen Psychiatry
22. Judd F, Mijch A. Depressive symptoms in patients with HIV infection.Aust NZ J Psychiatry
23. Perry S, Fishman B, Jacobsberg L, et al
. Relationships over 1 year between lymphocyte subsets and psychosocial variables among adults with infection by human immunodeficiency virus.Arch Gen Psychiatry
24. Perry SW, Jacobsberg LB, Fishman B, et al
. Psychological responses to serological testing for HIV.AIDS
25. Kern D, Collins M, Fultz T, et al
. An enhanced-sensitivity branched-DNA assay for quantification of HIV-1 RNA in plasma. J Clin Microbiol
26. Eldred L, Albert W, Chaisson R, Moore R. Adherence to antirretroviral drugs and pneumocystis prophylaxis in HIV disease.J Acquir Immune Defic Syndr Hum Retrovirol
27. Knobel H, Serrano C, Hernandez P, Pavesi M, Diez A. Acceptance, adherence and tolerance of antiretroviral therapy in HIV-infected patients. An Med Intern (Madrid)
28. Ickovics JR. Measures of adherence.Adherence to New HIV Therapies: A Research Conference.
Washington, DC. November 1997.
29. Norell S. Accuracy of patient interviews and estimates by clinical staff in determining medication compliance.Soc Sci Med
30. Ferrando S, Wall T, Batki S, et al
. Psychiatric morbidity, illicit drug use and adherence to zidovudine among injection drug users with HIV disease.Am J Drug Alcohol Abuse
31. Gerbert B, Maguire BT, Bleeker T, et al
. Primary care physicians and AIDS: attitudinal and structural barriers to care.JAMA
32. Fedder D. Drug use in the elderly: issues of non-compliance.Drug Intell Clin Pharm
33. Lipsitz J, Williams J, Rabkin J, et al
. Psychopathology in male and female intravenous drug users with and without HIV infection. Am J Psychiatry
34. Zorrilla E, McKay J, Luborsky L, Schmidt K. Relation of stressors and depressive symptoms to clinical progression of viral illness. Am J Psychiatry
35. Lyketsos C, Hoover D, Guccione M, et al
. Depressive symptoms as predictors of medical outcomes of HIV infection: Multicenter AIDS cohort study.
JAMA 1993, 270
36. Burack J, Barrett D, Stall R, et al
. Depressive symptoms and CD4 lymphocyte decline among HIV-infected men. JAMA
37. Markowitz JC, Rabkin JG, Perry SW. Treating depression in HIV-positive patients.AIDS
38. Mayne T, Vittinghoff E, Chesney M, et al
. Depressive affect and survival among gay and bisexual men infected with HIV.Arch Gen Psychiatry
39. Ohmit S, Schuman P, Schoenbaum E, et al
. Adherence to antiretroviral therapy (ART) among women in the HIV Epidemiology Research Study (HERS) and Women‚s Inter-Agency HIV Study (WIHS).12th World AIDS Conference,
Geneva, Switzerland, 1998 [Abstract 32347].
40. Lyketsos C, Hoover D, Guccione M, et al
. Changes in depressive symptoms as AIDS develops.Am J Psychiatry
41. Auerbach J. The role of behavioral research in HIV/AIDS prevention.Curr Op Infect Dis
42. Sweeney K, Mc Auley D, Pereira D. Personal significance: the third dimension.Lancet
43. Anonymous. AIDS Surveillance in Madrid.Boletin Epidemiológico de la Comunidad de Madrid,
44. Chesney M. New Antiretroviral therapies: adherence challenges and strategies.Antiviral Academy Meeting.
Mont Tremblant, Canada. March 1998.
45. Duffy D. Dialogue: the core clinical skill.Ann Intern Med