Rate and predictors of self-chosen drug discontinuations in highly active antiretroviral therapy-treated hiv-positive individuals

AIDS PATIENT CARE and STDsVolume 23, Number 1, 2009ª Mary Ann Liebert, Inc.
DOI: 10.1089=apc.2007.0248 Rate and Predictors of Self-Chosen Drug Discontinuations in Highly Active Antiretroviral Therapy-Treated Rita Murri, M.D.,1 Giovanni Guaraldi, M.D.,2 Piergiorgio Lupoli, Ph.D.,3 Raffaella Crisafulli, Ph.D.,3 Simone Marcotullio, Ph.D.,4 Filippo von Schloesser,4 and Albert W. Wu, Ph.D.5 Despite the clinical benefits of highly active antiretroviral therapy (HAART), sustained treatment remains a greatchallenge for HIV-infected people. The rate, consequences, and correlates of self-elected treatment interruptions(TI) are not known. The objectives of the study were to assess the rate of patient-elected TI in a cohort of HIV-infected people taking HAART, to evaluate whether patient-elected TI is correlated with suboptimal non-adherence, and to identify the predictors of self-chosen HAART interruptions. Using a Web-based cross-sectionalsurvey beginning in January 2006 primary outcomes were: (1) reports of having asked their physician to interruptthe current regimen (AskDisc) and (2) reports of at least one interruption of a minimum of 1 day of any of the drugsincluded in the regimen (INTERR). Three hundred fifty-nine people were enrolled; 296 were taking HAART.
Twenty-three percent self-reported suboptimal adherence, 45% reported AskDisc, and 25% INTERR. Forty per-cent of people reporting INTERR self-reported suboptimal adherence. As expected, AskDisc and INTERR werecorrelated with suboptimal adherence. The AskDisc group had higher CD4 cell counts and HIV RNA, moresymptoms, and took more convenient regimens. The INTERR group had higher HIV RNA, were more likely tosmoke, seek more information on HIV=AIDS, and less likely to take non-nucleoside reverse transcriptase inhib-itors (NNRTIs). The rate of self-chosen TI was high and often related to suboptimal adherence. These findings mayhelp clinicians to better monitor patients, and identify patients for targeted counseling.
suggests disadvantages to this strategy.3 Particularly, an in-creased risk of cardiovascular diseases was observed dur- Despiteimportantimprovementsintheconvenienceof ing the TI period.4 Similar results were confirmed in another antiretroviral regimens, long-term therapy is a major large study even showing that the heightened risks linked challenge for people living with HIV=AIDS (PLWHA). Highly to TI were not reversible after continuous treatment was re- active antiretroviral therapy (HAART) is now better tolerated; few daily pills and once-a-day regimens are common.1,2 Anecdotal evidence suggests that patients may undertake However, in the absence of alternative immunologic and TI, sometimes referred to as drug holidays, and not discuss vaccines strategies, HAART remains a life-long therapy and this decision with their physician. Few studies have exam- treatment fatigue is a key barrier to an optimal adherence to ined the issue of self-elected TI.5,6 The rate and predictors of patient-elected TI, particularly of when not agreed upon with Several studies have addressed the issue of treatment in- terruptions (TI) as a possible strategy to offset treatment fa- The objectives of the present study were to assess the rate of tigue and enhance quality of life, limit adverse events, reduce self-chosen TI in a cohort of HIV-infected people taking costs, and contain the emergence of multidrug-resistant virus.
HAART and identify the warning signs (i.e., predictors) of Results have been contradictory, but a recent large study 1Catholic University of Rome, Rome, Italy.
2University of Modena and Reggio Emilia, Modena, Italy.
4Fondazione Nadir Onlus, Rome, Italy.
5Department of Health Policy and Management, Bloomberg School of Public Health, The Johns Hopkins University, Baltimore, Maryland.
dence intervals (CI) were estimated. All analyses were donewith SPSS version 13.0 (SPSS, Chicago, IL).
This cross-sectional survey was conducted in conjunction with LONGIS (LONGitudinal Information Study), a pro- spective cohort study designed and supported by Nadir On-lus Foundation, a not-for-profit patient-based foundation, Between January and December 2006, 359 patients were enrolling adults HIV-infected people via the Web. The aim of the cohort is to explore, using a patient-centered approach, One hundred thirty-three subjects (44.9%) reported that preferences regarding antiretroviral therapy, belief in the ef- they had asked their physician to discontinue therapy in the 4 ficacy of HAART, adherence to drugs, the role of the patient in choosing or switching the HAART regimens, cotherapies, and Of the 296 patients taking HAART, 68 (23%) self-reported different aspects of the patient–physician relationship.
suboptimal adherence in response to the question: ‘‘How do In the present study patients were recruited through a you think you are taking therapy?’’ and 13 patients (5.7%) questionnaire posted on the Italian website of Nadir Onlus reported that they missed at least one dose per week, even if Foundation (www.nadironlus.org) between January and De-cember 2006. No specific selection procedures were adopted.
Visitors to the Nadir website were invited to participate in the Table 1. Characteristics of the Three Hundred survey; they could be directly linked to the questionnaire. No particular emphasis was placed on adherence to antiretrovirals or to satisfaction with therapy but only to patient preferences.
The following definitions of self-reported adherence were used: (1) patient rating report of accuracy of taking antiretro- viral therapy; (2) report of missing doses in a fixed period (how many doses the patient missed in the previous week and how many in the previous month), (3) reported timing of therapy (how often the patient took pills 2 hours before or after the prescribed time), and (4) reporting at least one interruption of a minimum of 1 day of any of the drugs included in the regi- men without having informed the physician either before or after (‘‘Have you ever discontinued your current regimen for at least 1 day without informing your physician, before or after the fact?’’). Suboptimal adherence was defined as answering ‘‘very bad,’’ ‘‘bad,’ or ‘‘not well enough’’ to the question: ‘‘How do you think you are taking therapy?’’ or those reporting having missed at least one dose in the previous week.
We also included a question on the willingness to dis- continue drugs (‘‘Have you ever asked your physician to discontinue your current regimen for a period?’’ with possible options not or yes). This latter was considered the main out- The survey also included questions on health status (‘‘How do you define your physical health?’’ and ‘‘How do you define your mental health?’’ with possible options ‘‘very bad,’ ‘ bad,’ ‘‘not well enough,’’ ‘‘good,’ or ‘‘excellent’ ) and on self-reported symptoms (25 among the most experienced symptoms in HIV- infected people taking HAART with possible options from ‘‘not at all’ to ‘‘very much’’). A symptom score was calculated summing scores for any single symptom. Data on age, gender, mode of HIV transmission, educational attainment, smoking, drinking alcohol, coinfection with viral hepatitis, as well as on therapy characteristics (type of drugs, number of doses, number of previous antiretroviral schemes), most recent HIV- RNA level, and most recent CD4 cell count were collected.
Bivariate analysis was performed to assess the correlation between the adherence dimensions and outcome and to identify other potential predictors (both subjective and ob- jective) associated with asking to discontinue drugs. Back- Smokers were defined those reporting to smoke any number ward stepwise logistic regression analysis was used to assess bDaily alcohol drinkers were defined those answering ‘‘yes, daily’’ the independent effects of the significant ( p < 0.1) explanatory at the question: ‘‘How often do you drink alcohol?’ variables on the outcome. Odds ratios (OR) and 95% confi- HAART, highly active antiretroviral therapy; SD, standard deviation; Table 2. Correlation of Different Adherence Definitions and of Drug Discontinuations with HIV RNA Adherence dimensions—Univariate OR (95%CI) they reported optimal adherence. Seventy-three patients people reported a repeated drug holiday (defined as stopping (24.7%) reported that they had discontinued therapy for at the regimen entirely for more than 48 hours),6 while in a Swiss least 1 day without informing their physicians, either before cohort in which a drug holiday was defined as missing all the or after the fact, and 40% self-reported suboptimal adherence.
drugs for at least 24 hours, it was reported only in 5.8%.7 Only 49 patients of 133 (36.8%) who had asked their physician The results of this study also suggest that these behaviors to discontinue therapy in the 4 weeks before the survey also are potential markers of suboptimal adherence. People who reported having discontinued their current regimen. One ask to discontinue therapy have a fourfold higher risk to miss hundred seven patients (36.1%) reported that they usually at least one dose in the previous week. We believe that in- took their medications 2 hours before or after the prescribed vestigating multiple aspects of adherence behavior, including time. Sixty-three percent of people had HIV RNA less than 50 timing of therapy and willingness of discontinuation of drugs, copies per milliliter and 11% between 50 and 500 copies per may allow better identification of people who need a stronger milliliter. Only 2.5% and 8.3% of enrolled people did not re- or a more targeted support for maintaining an optimal ad- port data on CD4 cell count or on HIV RNA, respectively. The herence and to prevent future nonadherent behavior. For risk of having an HIV RNA greater than 500 copies per mil- example, due to the long half-life of NNRTIs (namely efavir- liliter was higher for those reporting suboptimal adherence as enz and nevirapine), discontinuing this class of drugs at the well as for people asking to discontinue therapy or having same time of NRTI may lead to a period of NNRTI mono- discontinued therapy for at least 1 day (Table 2).
therapy with dangerous consequences on the selection of Of note, nonadherence with the prescribed dosing time was drug-resistant viruses. It is also possible that increasing and more frequent for people who asked to discontinue drugs (OR supporting an optimal adherence could prevent unplanned 4.62 [95% CI 2.77–7.68] compared to people who did not ask and casual TI.8,9 Appropriate counseling on the consequences to discontinue drugs) and for people who had discontinued of drug discontinuation, especially when not physician- drugs for at least 1 day (OR 3.56 [95% CI 2.06–6.16] compared driven and lasting for several days, could be crucial to moti- to people not having discontinued drugs). Asking to dis- vate patients to adhere to therapy or even, if necessary, to continue drugs was significantly correlated with reports of educate the patient ‘‘to be nonadherent in a rational way.’’10 having missed at least one dose in the previous week (OR 3.99 We also found that people taking NNRTIs were signifi- [95% CI 1.96–8.11] compared to people not asking to dis- cantly less likely to have discontinued drugs. Possible expla- nations of this result are (1) a selection bias (more adherent In Table 3, bivariate and multivariable analyses of factors people were prescribed NNRTI more frequently); (2) people associated with asking to discontinue therapy or to having taking NNRTI were more informed on the importance to discontinued drugs are shown. People who asked to dis- avoid an uncontrolled discontinuation of the drug due to the continue drugs had higher CD4 cell counts, higher HIV RNA, long half-life; (3) NNRTI-containing regimens are better tol- and more symptoms, took more convenient regimens, and erated11,12 leading to less treatment fatigue; (4) people taking self-reported suboptimal adherence. People who reported NNRTI may have previously expressed concerns with prior, having discontinued drugs had higher HIV RNA, were more more complicated regimens resulting in a higher satisfaction likely to smoke, have suboptimal adherence, seek more in- with therapy with the current regimen.
formation on HIV=AIDS, and were less likely to take non- It has been demonstrated that drug-related symptoms are nucleoside reverse transcriptase inhibitors (NNRTIs).
related both to a higher rate of discontinuations13,14 but also toa higher risk of suboptimal adherence.15,16 Treatment fatiguemay be an important reason for willingness to discontinue HAART in people reporting higher symptom scores. At the In the present study, nearly half of those surveyed reported same time, people with better clinical status (higher CD4 cell having asked their physician to interrupt HAART, and nearly count) or on more convenient regimens (with few daily doses one quarter had interrupted HAART for at least 1 day without and pills) were paradoxically more likely to discontinue informing their physician. Reports of one of these behaviors drugs. This may be because patients who are perceived likely was significantly associated with suboptimal adherence and to have adherence problems may be prescribed simpler regi- mens. It is also likely that people with more complex regimens There have been few reports on patient treatment inter- are those with less available therapeutic options and these ruptions from traditional cohort studies of HIV-infected individuals are more aware of the importance of maintaining people taking HAART. In a French study, 27% of HIV-infected the current regimen. Moreover, people on more convenient regimens could underestimate the real complexity of the 5. Calmy A, Nguyen A, Montecucco F, et al. HIV activates regimen they are taking and consequently believe that they markers of cardiovascular risk in a randomized treatment are taking a less aggressive regimen or that they need less interruption trial: STACCATO [Abstract 10]. 15th Confer- treatment. It would be important to investigate in further ence on Retroviruses and Opportunistic Infections. Boston, studies whether a simpler or simplified HAART regimen may be associated with being healthier and hence make patients 6. Parienti JJ, Massari V, Descamps D, et al. Predictors of viro- more prone to missing doses or discontinuing them.
logic failure and resistance in HIV-infected patients treated There are several limitations to this study. First, the cross- with nevirapine- or efavirenz-based antiretroviral therapy.
sectional design did not allow us to establish the direction of Clin Infect Dis 2004;38:1311–1316.
relationships among the variables in the study. Second, all 7. Glass TR, De Geest S, Weber R, et al. Correlates of self- reported nonadherence to antiretroviral therapy in HIV- variables were self-reported by patients, including clinical infected patients: The Swiss HIV Cohort Study. J Acquir measures such as viroimmunologic parameters. This may Immune Defic Syndr 2006;41:385–392.
increase the variability beyond that in which these data are 8. Sidat M, Fairley C, Grierson J. Experiences and percep- measured objectively. However, it should be noted that the tions of patients with 100% to highly active antiretroviral rate of missing data was very low and the population char- therapy: A qualitative study. AIDS Patient Care STDs 2007; acteristics are similar to that of other cohorts of HIV-infected people. It can be also argued that self-report overestimates 9. Oyugi JH, Byakika-Tusiime J, Ragland K, et al. Treatment some outcomes such as adherence to therapy.10 A self- interruptions predict resistance in HIV-positive individuals reported web survey was chosen to avoid the bias of the purchasing fixed-dose combination antiretroviral therapy in presence of the physician in answering to the question on self- Kampala, Uganda. AIDS 2007;21:965–971.
chosen TI and potentially reduce this overestimation, even 10. O’Connor P. Improving medication adherence. Arch Intern though internet access may induce a bias in the sample.17 Third, some of the measures used in the present survey were 11. Chou R, Fu R, Huffman LH, Korthuis PT. Initial highly- not previously validated. Fourth, genotypic data were not active antiretroviral therapy with a protease inhibitor versus available in the case of virologic failure. Longitudinal studies a non-nucleoside reverse transcriptase inhibitor: Discrepan- are warranted to confirm the present findings.
cies between direct and indirect meta-analyses. Lancet 2006; In conclusion, the proportion of people asking for or un- dergoing self-elected TI appears to be high. TI can be con- 12. Riddler SA, Haubrich R, DiRienzo G, et al. A prospective, sidered a measure of suboptimal adherence. The willingness randomized, phase III trial of NRTI-, PI-, and NNRTI- and desire of PLWHA on HIV therapy to undergo drug holi- sparing regimens for initial treatment of HIV-1 infection: days due to treatment fatigue should be discussed in the ACTG 5142 [Abstract THLB0204]. Program and abstracts ofthe XVI International AIDS Conference. Toronto, Canada: context of the patient–physician relationship. Risks and un- certainties of monitored TI strategies may be different ac- 13. O’Brien ME, Clark RA, Besch CL, Myers L, Kissinger P.
cording to clinical status and to type of therapy and should be Patterns and correlates of discontinuation of the initial balanced with a higher risk of patient unilateral and danger- HAART regimen in an urban outpatient cohort. J Acquir ous decisions on discontinuation. Findings of the present Immune Defic Syndr 2003;34:407–414.
study may help to better monitor patients in clinical practice, 14. d’Arminio Monforte A, Cozzi Lepri AC, Rezza G, et al. In- interpret viroimmunologic results, prevent the appearance of sights into the reasons for discontinuation of the first highly drug resistance or progression of HIV disease and, ultimately, active antiretroviral therapy (HAART) regimen in a cohort identify patients who need targeted counseling.
of antiretroviral naı¨ve patients. I.CO.N.A. Study Group.
AIDS 2000;14:499–507.
15. Ammassari A, Murri R, Pezzotti P, et al. Self-reported symptoms and medication side effects influence adherence No competing financial interests exist.
to highly active antiretroviral therapy in persons with HIVinfection. J Acquir Immune Defic Syndr 2001;28:445–449.
16. Heath KV, Singer J, O’Shaughnessy MV, Montaner JS, Hogg RS. Intentional nonadherence due to adverse symptoms as- 1. DHHS Panel on Antiretroviral Guidelines for Adults and sociated with antiretroviral therapy. J Acquir Immune Defic Adolescents. Guidelines for the use of antiretroviral agents in HIV-1–infected adults and adolescents, October 2006.
17. Murri R, Marcotullio S, Lupoli P, von Schloesser F. Is ‘‘once- www.aidsinfo.nih.gov= (Last accessed July 2007).
daily’’ regimen a key strategy for improving adherence to 2. May MT, Sterne JA, Costagliola D, et al. HIV treatment re- antiretroviral regimens? J Acquir Immune Defic Syndr 2006; sponse and prognosis in Europe and North America in the first decade of highly active antiretroviral therapy: A col-laborative analysis. Lancet 2006;368:451–458.
3. Strategies for Management of Antiretroviral Therapy (SMART) Study Group; El-Sadr WM, Lundgren JD, Neaton JD, et al. CD4þ count-guided interruption of antiretroviral treatment. N Engl J Med 2006;355:2283–2296.
4. Ananworanich J, Gayet-Ageron A, Le Braz M, et al. CD4- guided scheduled treatment interruptions compared with continuous therapy for patients infected with HIV-1: Resultsof the Staccato randomised trial. Lancet 2006;368:459–465.

Source: http://www.edgeconsulting.it/publications/DiscontLongis%20AIDSPatCareSTD2009.pdf