Nonadherence to Highly Active Antiretroviral Therapy - Rega Institute

10 downloads 858 Views 149KB Size Report
A notebook was provided to detail any ... fusion with blood products in 9%. ... 93 (13.3). 64 (33.1). ,0.0001. Drug holidays (n/100 days). 0.8 (4.8). 0.0 (0.2).
AIDS RESEARCH AND HUMAN RETROVIRUSES Volume 18, Number 5, 2002, pp. 327–330 © Mary Ann Liebert, Inc.

Nonadherence to Highly Active Antiretroviral Therapy: Clinically Relevant Patient Categorization Based on Electronic Event Monitoring ERIC VAN WIJNGAERDEN, 1 VEERLE DE SAAR,2 VEERLE DE GRAEVE,2 ANNE-MIEKE VANDAMME, 3 KRISTIEN VAN VAERENBERGH, 3 HERMAN BOBBAERS,1 ANN DESCHAMPS,1 HELGA CEUNEN, 1 and SABINA DE GEEST2,4

ABSTRACT Adherence to highly active antiretroviral therapy (HAART) is crucial, but which aspects of drug-taking behavior are important remain largely unknown. In a prospective observational study, 43 HIV-1-infected patients taking HAART underwent electronic event monitoring (EEM). Taking adherence was defined as the percentage of doses taken compared with the number prescribed, dosing adherence was defined as the percentage of days on which all doses were taken, and timing adherence was defined as the percentage of doses taken within 1 hr of the time prescribed. Drug holidays were defined as periods of no drug intake for .24 hr. Cluster analysis, including the four EEM parameters, was used and refined to construct an algorithm to discriminate patients. Patients were categorized as nonadherent if they had a taking adherence of ,90%, or a dosing adherence of ,75% and at least 1 drug holiday, or a timing adherence of ,80% and at least 1 drug holiday, or .6 drug holidays per 100 days. All four EEM parameters differed significantly (p , 0.0001) between the two groups. Adherent patients had a better outcome, as shown by a larger drop in viral load (p 5 0.011) and rise in CD4 1 cell count (p 5 0.035), showing that the algorithm-based categorization is clinically relevant. INTRODUCTION

W

HER EVER AVAILAB LE ,

highly active antiretroviral therapy (HAART) has become the standard of care in the treatment of HIV-infected patients.1 Soon after its introduction, its profound impact on morbidity and mortality was appreciated.2,3 Drug failure due to viral resistance is a common and serious problem, limiting the long-term benefit from this treatment. It was soon realized that drug adherence plays a major role in avoiding drug failure.4,5 Despite the advances being made, HAART remains probably the most complex life-long drug regimen prescribed to a large and unselected patient population. Considerable short-term and an evolving spectrum of long-term toxicities complicate its use. It can thus be appreciated why adherence was labeled the “Achilles heel” of HAART. 4 Just how much adherence is sufficient, and which aspects in

the dynamics of drug-taking behavior are most important, are questions to which the answers remain largely unknown. We conducted a study using electronic event monitoring (EEM) to try to clarify these important issues.

MATERIALS AND METHODS Study population This was a single-center, prospective, observational study. All HIV-infected patients over 18 years old, taking triple or quadruple HAART including at least one protease inhibitor (PI) for at least 1 month, were eligible. Written informed consent was obtained. Inclusion was from September 1998 through December 1998. Exclusion criteria were inability to read and active opportunistic infection or malignancy.

1 Department

of Internal Medicine, University Hospitals of KU-Leuven, Leuven, Belgium. for Health Services and Nursing Research, KU-Leuven, Leuven, Belgium. Institute for Medical Research, KU-Leuven, Leuven, Belgium. 4 Institute of Nursing Science, University of Basel, Basel, Switzerland. 2 Center 3 Rega

327

328

VAN WIJNGAERDEN ET AL. TABLE 1.

ELECTRONIC EVENT MONITORING AND VIROLOGICA L OUTCOME PARAMETERS ADHERENT GROUP , AND THE NONADHERENT GROUP

IN

ALL PATIENTS ,

THE

Median (IQR) All (n 5 43) EEM parameter Taking adherence (%) Dosing adherence (%) Timing adherence (%) Drug holidays (n/100 days) Virological outcome D log viral load D CD41 cell count

98 92 86 0.8

(5.3) (18.0) (31.5) (4.8)

22.8 (2.4) 154 (122)b

Adherent (n 5 26) 99 96 93 0.0

(2.3) (9.3) (13.3) (0.2)

23.00 (1.95) 164 (107)b

Nonadherent (n 5 17) 90 75 64 6.2

(14.1) (34.0) (33.1) (6.9)

p Valuea ,0.0001 ,0.0001 ,0.0001 ,0.0001

21.80 (2.97) 118 (125)b

0.011 0.035

a Wilcoxon two-sample test for the comparison between adherent and nonadherent patients, except Student t test for comparison of D log viral load. b Mean (SD).

Assessment of adherence

Outcome parameters

Adherence was monitored for a 3- to 4-month period using EEM (Edem; AARDEX, Zug, Switzerland). The system consists of a medication bottle, with a cap containing a microelectronic circuit registering day and time of each opening. Patients were thoroughly instructed on the use of the system. One drug was monitored, preferentially a PI. In four patients taking ritonavir liquid formulation as the sole PI, a reverse transcriptase inhibitor was monitored. A notebook was provided to detail any deviation from the protocol. We adopted a strategy in which patients who preferred not to use the EEM system outside of their home kept a record of the times of medication, which were afterward merged with the EEM data. Four operational definitions of adherence were assessed. Taking adherence was defined as the percentage of doses taken in comparison with the number of doses prescribed. Dosing adherence was defined as the percentage of days in which the correct number of doses was taken. Timing adherence was defined as the percentage of doses taken within 1 hr before and after the time prescribed. Drug holidays were defined as periods of no drug intake for .24 hr and expressed per 100 days. Classification of nonadherence and adherence with HAART was obtained in a three-step fashion. In a first step patients were categorized by degree of adherence, using a multivariate technique, that is, iterative partitioning methods of cluster analysis, including the EEM parameters. This was complemented by univariate analysis (ANOVA) to verify whether clusters indeed differed significantly on the variables entered in the cluster analysis (using SPSS 9.0; SPSS, Chicago, IL). Because the assumptions underlying cluster analysis were only partially fulfilled, we regarded cluster analysis only as a first step in discriminating the patients in two groups. During the second step a table including all EEM parameters was carefully scrutinized in order to find a cutoff that discriminated the patient in adherence profile.6 The membership of the initial cluster solution was also taken into consideration during this process. In a last step post-hoc validation of the algorithm by comparing groups on outcome parameters was done in order to assess the clinical meaningfulness of the categorization.

Viral load was assessed by the Quantiplex HIV-1 RNA 3.0 test (Chiron, Emeryville, CA). CD41 cell count was determined by flow cytometry (Becton Dickinson, Franklin Lakes, NJ). The reduction in viral load and the rise in CD41 cell count were calculated from the start of HAART to the end of the monitoring period.

RESULTS Sixty-eight patients were eligible, 13 refused participation, 7 patients terminated prematurely, and 5 lost the EEM device. The final study sample consisted of 43 patients. Their mean age was 42 6 8.5 years (mean 6 SD), 84% were male, and 86% were white. HIV transmission was by heterosexual contact in 47%, by homosexual contact in 44%, and unknown or by transfusion with blood products in 9%. The median CD41 cell count at the start of the monitoring period was 418 cells/ml (interquartile range [IQR] 243 cells/ml) and the median viral load was ,50 copies/ml (IQR 149 copies/ml). The characteristics of the 25 patients not included (data not shown) were not different from those participating, except that they were younger (36 6 9.9 years, Student t test p 5 0.005). EEM was performed for a mean 112 6 24 days. The PI monitored was indinavir (three times daily) in 17, nelfinavir (three times daily) in 16, and saquinavir (with ritonavir) twice daily

TABLE 2. ALGORITHM THAT MAXIMALLY DISCRIMINATES NONADHERENT FROM A DHERENT PATIENTS Taking adherence ,90% OR Dosing adherence ,75% and at least 1 drug holiday OR Timing adherence ,80% and at least 1 drug holiday OR Number of drug holidays .6/100 days

SUBCLINICAL NONADHERENCE TO HAART in 6. In four patients taking a ritonavir liquid formulation, a reverse transcriptase inhibitor (RTI) taken twice daily was monitored. HAART was taken for a median duration of 16 months (IQR 14 months) at the start of EEM monitoring. Adherence data are shown in Table 1. Taking adherence was high at 98% and the median number of drug holidays was less than 1 per 100 days. On the basis of the initial two-group cluster solution, nine patients were classified as nonadherent. The algorithm (Table 2) subsequently developed to maximally discriminate patients on the basis of all four adherence parameters resulted in 26 patients considered adherent and 17 patients considered subclinically nonadherent. Subclinical nonadherence refers to situations not yet accompanied by, but potentially leading to, a clinical event or poor clinical outcome. Table 1 lists the adherence data of both groups, which as expected discriminated both groups significantly on all four EEM parameters (p , 0.0001). Drug holidays were present in all nonadherent patients with a median of 6.2 per 100 days, whereas only six adherent patients (23%) had a drug holiday. A post-hoc comparison of outcome in both groups was performed, showing that the viral load reduction in adherent patients was more pronounced and the mean rise in CD41 cell count was more important (Table 1).

CONCLUSION By using EEM monitoring and four operational definitions of adherence, we developed an algorithm that was capable of categorizing two groups of adherent and subclinically nonadherent patients in our study population. Overall, 40% of patients were subclinically nonadherent as defined in this way. The study group was small, the treatment history heterogeneous, preexisting drug resistance mutations were common (data not shown), and the composition and duration of the current HAART regimens were variable. Electronic event monitoring was used for a 3- to 4-month period only to determine the level of adherence. Past and current adherence have been found to be the most reliable predictors of future adherence with medication regimens.7 We believe that the 3 to 4 months of EEM monitoring provided a reliable estimate of the patients’ overall level of adherence during the whole period in which outcome while undergoing the current HAART regimen was assessed. Despite these problems, it could be shown that these groups differed significantly in their clinical outcome as measured by both viral load reduction and rise in CD41 cell count. This shows that the algorithm-based division is clinically relevant and may be prognostic for long-term outcome. Several questions remain, however. Which dimension(s) of medication-taking dynamics are most important for effectiveness of HAART? Is it actual medication intake irrespective of timing, regularity of medication taking, adherence to meal prescriptions (this aspect cannot be assessed by EEM), or major drug lapses that will jeopardize HAART? Drug holidays seem to play a major role, as was suggested early on in the history of HAART. 8 One may argue that regularity of medication intake is crucial for some drugs for which trough levels come close to inhibitory concentrations (e.g., indinavir) but not if this is not the case. Yet, deviation from timing is generally consid-

329 ered to be of less importance than not taking the drug at all (“better late than never”) and may be the first sign of a more relaxed (but unfortunately nonadherent) attitude toward taking medication. Timing adherence emerges as a valuable EEM parameter in addition to drug holidays to define deviation from the dosing schedule, as it combines both intake and regularity dimensions of medication dynamics and is thus more sensitive to track deviating medication dynamics than taking or dosing adherence. Can our findings be generalized to other patient populations? No intravenous drug users were represented in our study population, unlike in most populations in the Western world. The taking adherence in the group as a whole was exceptionally high as compared with other studies using EEM, 9,10 which may influence cutoff levels. Validation in another patient population would be an important next step. Furthermore, drug regimens evolve rapidly. Most patients in our study were undergoing three-times-daily PI regimens, whereas at present most are taking PIs twice daily. It can be expected that the effects of nonadherence are not equal for all drugs. Emergence of resistance is quicker for some drugs (e.g., nonnucleoside reverse transcriptase inhibitors) than others. This means that the short-term clinical relevance of nonadherence will probably be more dramatic for the former than for the latter. It may be that clinically relevant drug adherence needs to be defined differently for different drugs or drug combinations. Nevertheless, we think the algorithm we developed can be useful both in further research on adherence and as a tool to categorize adherence in clinical trials of HAART.

ACKNOWLEDGMENT This work was funded by a research grant from Glaxo Wellcome Belgium.

REFERENCES 1. Carpenter CC, Cooper DA, Fischl MA, et al.: Antiretroviral therapy in adults: Updated recommendations of the International AIDS Society—USA panel. JAMA 2000;283:381–390. 2. Palella FJ, Delaney KM, Moorman AC, et al.: Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med 1998;338:853–860. 3. Staszewski S, Demasi R, Hill AM, and Dawson D: HIV-1 RNA, CD4 cell count and the risk of progression to AIDS and death during treatment with HIV-1 reverse transcriptase inhibitors. AIDS 1998;12:1991–1997. 4. Altice FL and Friedland GH: The era of adherence to HIV therapy. Ann Intern Med 1998;129:503–505. 5. Paterson DL, Swindells S, Mohr J, et al.: Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med 2000;133:21–30. 6. De Klerk E, Van der Linden S, van der Heijde D, and Urquhart J: Facilitated analysis of data on drug regimen compliance. Stat Med 1997;16:1653–1664. 7. Dunbar-Jacob J, Erlen JA, Schlenk EA, Ryan M, Sereika SM, and Doswell WM: Adherence in chronic disease. Annu Rev Nurs Res 2000;18:48–90. 8. Vanhove GF, Schapiro JM, Winters MA, Merigan TC, and

330 Blaschke TF: Patient compliance and drug failure in protease inhibitor monotherapy. JAMA 1996;276:1955–1956. 9. Kastrissios H, Suarez JR, Katzenstein D, Girard P, Sheiner LB, and Blaschke TF: Characterizing patterns of drug-taking behavior with a multiple drug regimen in an AIDS clinical trial. AIDS 1998;12: 2295–2303. 10. Golin C, Liu H, Hays R, et al.: Self-reported adherence to protease inhibitors substantially overestimates on objective measure. In: 6th Conference on Retroviruses and Opportunistic Infections, Chicago, Illinois, January 31–February 4, 1999. [Abstract 95]

VAN WIJNGAERDEN ET AL. Address reprint requests to: Eric Van Wijngaerden Department of Internal Medicine University Hospitals of KU-Leuven Herestraat 49 3000 Leuven, Belgium

E-mail: [email protected]

This article has been cited by: 1. Arne Christensen, Lona Louring Christrup, Paul Erik Fabricius, Marzena Chrostowska, Michal Wronka, Krzysztof Narkiewicz, Ebba Holme Hansen. 2010. The impact of an electronic monitoring and reminder device on patient compliance with antihypertensive therapy: a randomized controlled trial. Journal of Hypertension 28:1, 194-200. [CrossRef] 2. Michael Horberg , Michael Silverberg , Leo Hurley , Gerald Delorenze , Charles Quesenberry . 2008. Influence of Prior Antiretroviral Experience on Adherence and Responses to New Highly Active Antiretroviral Therapy RegimensInfluence of Prior Antiretroviral Experience on Adherence and Responses to New Highly Active Antiretroviral Therapy Regimens. AIDS Patient Care and STDs 22:4, 301-312. [Abstract] [PDF] [PDF Plus] 3. R. Scott Braithwaite, Steven Shechter, Chung-Chou H. Chang, Andrew Schaefer, Mark S. Roberts. 2007. Estimating the Rate of Accumulating Drug Resistance Mutations in the HIV Genome. Value in Health 10:3, 204-213. [CrossRef] 4. Honghu Liu, Loren G. Miller, Carol E. Golin, Ron D. Hays, Tongtong Wu, Neil S. Wenger, Andrew H. Kaplan. 2007. Repeated measures analyses of dose timing of antiretroviral medication and its relationship to HIV virologic outcomes. Statistics in Medicine 26:5, 991-1007. [CrossRef] 5. Kevin D. Dieckhaus , Victoria Odesina . 2007. Outcomes of a Multifaceted Medication Adherence Intervention for HIV-Positive PatientsOutcomes of a Multifaceted Medication Adherence Intervention for HIV-Positive Patients. AIDS Patient Care and STDs 21:2, 81-91. [Abstract] [PDF] [PDF Plus] 6. Karina M Berg, Julia H Arnsten. 2007. Practical and Conceptual Challenges in Measuring Antiretroviral Adherence. JAIDS Journal of Acquired Immune Deficiency Syndromes 43:Supplement 1, S79-S87. [CrossRef] 7. Sabina De Geest, Petra Schafer-Keller, Kris Denhaerynck, Nicole Thannberger, Susanne Kofer, Andreas Bock, Christian Surber, Jurg Steiger. 2006. Supporting medication adherence in renal transplantation (SMART): a pilot RCT to improve adherence to immunosuppressive regimens. Clinical Transplantation 20:3, 359-368. [CrossRef] 8. Honghu Liu, Loren G. Miller, Ron D. Hays, Carol E. Golin, Tongtong Wu, Neil S. Wenger, Andrew H. Kaplan. 2006. Repeated Measures Longitudinal Analyses of HIV Virologic Response as a Function of Percent Adherence, Dose Timing, Genotypic Sensitivity, and Other Factors. JAIDS Journal of Acquired Immune Deficiency Syndromes 41:3, 315-322. [CrossRef] 9. Andrew J. Levine, Charles H. Hinkin, Steven A. Castellon, Karen I. Mason, Mona N. Lam, Adam Perkins, Marta Robinet, Douglas Longshore, Thomas Newton, Hector Myers, Ramani S. Durvasula, David J. Hardy. 2005. Variations in Patterns of Highly Active Antiretroviral Therapy (HAART) Adherence. AIDS and Behavior 9:3, 355-362. [CrossRef] 10. Ann E. Deschamps , Veerle De Graeve , Eric Van Wijngaerden , Veerle De Saar , Anne-Mieke Vandamme , Kristien Van Vaerenbergh , Helga Ceunen , Herman Bobbaers , Willy E. Peetermans , Peter J. De Vleeschouwer , Sabina De Geest . 2004. Prevalence and Correlates of Nonadherence to Antiretroviral Therapy in a Population of HIV Patients Using Medication Event Monitoring System®Prevalence and Correlates of Nonadherence to Antiretroviral Therapy in a Population of HIV Patients Using Medication Event Monitoring System®. AIDS Patient Care and STDs 18:11, 644-657. [Abstract] [PDF] [PDF Plus] 11. Zelee Hill , Carl Kendall , Manuel Fernandez . 2003. Patterns of Adherence to Antiretrovirals: Why Adherence Has No Simple MeasurePatterns of Adherence to Antiretrovirals: Why Adherence Has No Simple Measure. AIDS Patient Care and STDs 17:10, 519-525. [Abstract] [PDF] [PDF Plus]