Kunutsor, Setor MBChB, MSt*; Evans, Morgan MRCP*; Thoulass, Janine MRCP*; Walley, John MBBS, DTM&H, MCOMMH, FFPH*; Katabira, Elly FRCP†; Newell, James N MA, MSc, PhD*; Muchuro, Simon MBChB, MPH†; Balidawa, Hudson MD, MMed, DTMPH, MPH‡; Namagala, Elizabeth MBChB, MMed‡; Ikoona, Eric MBChB, MMed, MPH‡
From the *Nuffield Centre for International Health and Development, Institute of Health Sciences, Leeds University, Leeds, United Kingdom; †Department of Medicine, Makerere University College of Health Sciences, Kampala, Uganda; and ‡Uganda National Aids Control Program, Uganda.
Received for publication October 17, 2009; accepted January 6, 2010.
The project was funded by the Communicable Disease Research Program Consortium led by the Nuffield Centre for International Health and Development at Leeds University which itself is funded by the Department for International Development, United Kingdom.
Correspondence to: Setor Kunutsor, MBChB, MSt, Malaria Consortium, COMDIS MUK, PO Box 8045, Kampala, Uganda. (e-mail: firstname.lastname@example.org).
Background: Many antiretroviral treatment (ART) adherence measurement methods have been employed by different studies, but no single method has been found to be appropriate for all settings. This study aimed to determine baseline levels of adherence using 2 measures of adherence.
Methods: Levels of adherence in 967 patients continuing to receive ART in 4 health facilities were assessed over a 28-week period using a clinic-based pill count method and a patient self-report questionnaire. Factors associated with adherence were also determined.
Results: Mean adherence (95% confidence interval) was 97.3% (96.8% to 97.9%) and 98.4% (97.9% to 98.8%) for the clinic-based pill count and patient self-report methods, respectively. Proportion of clients achieving optimal adherence (≥95%) was 89.9% by pill count and 94.2% by self-report. The 2 adherence measures were closely correlated with each other (r = 0.87, P = 0.000). Adherence increased with age (P = 0.014) with patients aged 40 years and below being less likely to achieve optimal adherence [odds ratio = 0.55; 95% confidence interval (0.34 to 0.89)].
Conclusions: There is a very high level of optimal adherence among patients still on treatment. The combined use of these 2 replicable and reliable methods of measuring adherence is vital to ART programs in resource-constrained settings.
At the end of 2007, an estimated 33 million people were living with HIV worldwide; the majority of these in the developing world, with approximately 22 million of these in sub-Saharan Africa.1 HIV/AIDS is a serious public health problem in Uganda. In the early 1990s, HIV prevalence peaked in Uganda at 15% among adults, and more than 30% among pregnant women in the cities.2 After more than 2 decades of the HIV/AIDS epidemic, after vigorous public education campaigns, prevalence is now estimated at 5.4% amongst adults.1 Currently, more than 350,000 HIV-positive individuals are clinically eligible for antiretroviral treatment (ART),3 and access to ART is being scaled up throughout the country since the roll-out of free ART in 2004.
With increasing access to ART, it is important to find efficient, feasible, and replicable methods of maintaining high levels of adherence. Assessing adherence is vital in any ART program, for both clinicians and researchers. However, there is no gold standard approach, and the best way to measure it must therefore be defined according to context.4
This study, part of a 3-phase study to identify feasible and replicable methods of improving adherence for use in the context of resource-constrained settings, set out to use a multimethod approach to ascertain baseline levels of adherence and assess factors associated with optimal adherence to ART.
The study was conducted in 4 districts in Uganda namely Kayunga, Jinja, Lyantonde, and Masaka. Uganda with Kampala as its capital is located in the great lakes region of East Africa.
Study Design and Participants
The study was based on a prospective cohort design. The study population included adults (18 years and older) who were receiving treatment at the 4 health care facilities. All patients visiting the sites during the study period, who were in this age range, and on ART were included. Study participants were followed up between December 2008 and June 2009. The ethics committee of the Makerere University Medical School reviewed and approved the study procedures and data collection instruments.
Adherence levels of 95% and over were categorized as “optimal,” and levels less than 95% were considered to be “suboptimal”. The cut-off points for adherence levels used in this study are the ones being implemented by the Uganda AIDS Control Program. The primary measure of adherence at a facility was the proportion of patients achieving 95% adherence.
ART adherence was assessed approximately every 4 weeks for 28 weeks using 2 complementary adherence measurement tools: a clinic-based pill count method and a patient self-report questionnaire. Adherence using pill count was calculated by dividing the number of pills actually taken by the number of pills the client was required to take over the reporting period multiplied by 100. Majority of patients were taking the first-line fixed-dose combination regimen for Uganda consisting of zidovudine, lamivudine, and nevirapine, and the alternative first-line fixed-dose combination consisting of stavudine, lamivudine, and nevirapine. All antiretroviral pills were counted together, and adherence calculated for the regimen as a whole for those not on fixed-dose combinations. In the self-report questionnaire, clients were asked about the number of pills missed in the last month and if they had missed any pills for 3 consecutive days in the last month.
Data for the adherence tools were entered into an MS Access database and SPSS version 15.0 for windows and analysis conducted using the same packages. Adherence of >100% (due to ingesting more pills than prescribed) was censored to 100%. The mean adherence for the study sample and proportions of clients achieving adherence levels of 95% and over were determined and 95% confidence intervals (CIs) calculated. Individual mean adherence levels were used to determine the association between the 2 measures of adherence using Pearson correlation coefficient. Factors associated with adherence were determined using univariate and multivariate analyses. The P values <0.05 (2-sided) were considered statistically significant.
Demographic Data and Patient Characteristics
A total of 1590 clients from the 4 sites were followed up, but results of only 967 were analyzed. The rest were excluded because of reasons such as loss to follow-up, deaths, and transfer outs (Table 1).
Adherence Levels and Measurement Tools
Mean adherence was 97.3% (95% CI: 96.8% to 97.9%) for pill count and 98.4% (95% CI: 97.9% to 98.8%) for patient recall. Assessments of adherence to ART using the 2 measurement tools are summarized in Table 2. A Pearson correlation showed the 2 adherence measures to be closely and significantly correlated with each other (r = 0.87, P = 0.000) (Table 2).
Predictors of Adherence
Results from univariate and multivariate analyses are reported (Tables 3 and 4). Table 3 shows proportions of clients achieving optimal adherence by adherence measurement tool and their various demographic and treatment characteristics.
Linear regression analysis showed that adherence was significantly and positively related to age (β = 0.079, P = 0.014). There was no significant association with duration on antiretrovirals (ARVs). Results from a multiple linear regression analyses showed that for a given duration on ARVs, adherence increased by 0.8% per decade. A multivariate logistic regression analysis showed that those in age category 40 years and younger were less likely to achieve optimal adherence (Table 4).
Major reasons cited for suboptimal adherence were inaccessibility to medication refills because of transportation costs and financial constraints (62.1%) and forgetfulness (21.6%). Other reasons included stigmatization and disclosure issues (6.1%); being too ill to take medication (1.5%); and adverse effects of treatment regimen such as diarrhoea, vomiting, and itchy body rash (0.8%).
Our findings show that patients, who continue on ART treatment in this resource-limited setting, take on average 98% of their prescribed ART. This is very encouraging and provides an excellent basis to continue the roll out of ART to a greater number of people in need. The adherence levels are similar for the more subjective patient recall and the more objective pill count measures of adherence. These high levels of adherence are broadly consistent with reports from other resource-limited settings in Africa using the clinic-based pill count and self-report measures.5-7
Adherence Measurement Tools
Methods for measuring adherence to ART in clinical practice should be reliable, economical, and replicable in resource-limited settings. There was a close and significant correlation between the 2 adherence measures used, which was also demonstrated by Oyugi et al8 in their study of multiple validated measures of adherence. In our study, there were slightly lower levels of adherence when adherence was measured by the more objective pill count measure, which is expected.
Our study, together with other studies, suggests that it is reasonable to use both methods in combination in routine clinical practice because they are both replicable and feasible in resource-constrained settings. Indeed, Vitolins et al4 in their study reported that the use of 2 or more measures of adherence allowed for the strength of one method to compensate for the limitations of the other and to more accurately capture the information needed to determine adherence levels.
Factors Associated With Adherence
Some studies have found significant associations between good adherence and male gender and older age,9-11 but other studies have found increasing age in years as significantly associated with suboptimal adherence.12,13 In our study, adherence increased with age, with people 40 years and below being less likely to achieve ≥95% adherence levels. The reason for this is not clear, but may be due to lower economic status14 or fear of social rejection or stigmatization when their HIV status is known. Generally, the role of social and demographic characteristics such as race, gender, age, and educational level as predictors of adherence has produced mostly inconsistent results,13,15 and this will reflect local customs and industry.
Despite the fact that ARV drugs are supplied free, transportation costs and financial constraints were the main reasons given for suboptimal adherence in this study. Other studies have also reported this finding.16 Financial constraints with its possible consequence of suboptimal adherence is a cause for concern. Poverty does present significant barriers to provision of a comprehensive HIV/AIDS program and provision of ART. There is the need to develop practical and sustainable ways of providing ART such as the home-based AIDS care program, which has been shown to promote good adherence and response to ART in resource-constrained rural African settings.17
Our study was potentially subject to biases. We were not able to corroborate adherence measurements with immunological, virological, or clinical outcomes because of financial constraints particularly the cost of additional laboratory monitoring in this setting. The data were derived from just 4 sites, in the south and west of Uganda, and so may not be generalizable to the whole country.
The study shows that high levels of adherence to antiretroviral therapy can be achieved in resource-constrained areas in sub-Saharan Africa. These high levels of adherence were in those patients remaining on treatment, and there is still the need to identify and implement adherence support interventions to retain clients on ART to maintain these high levels of adherence. Patient self-report and pill counts are a reliable and feasible combination of measures to monitor adherence in resource-constrained settings, such as our study sites.
We acknowledge the support of the Ugandan AIDS Control Programme in facilitating this study. The study team also acknowledges the contributions of the adherence workers and data collectors at the various sites.
2. Stoneburner RL, Low-Beer D. Population-level HIV declines and behavioural risk avoidance in Uganda
4. Vitolins MZ, Rand CS, Rapp SR, et al. Measuring adherence to behavioural and medical interventions. Control Clin Trials
5. Laurent C, Diakhate` N, Fatou Ngom Gueye N, et al. The Senegalese government's highly active antiretroviral therapy initiative: an 18-month follow-up study. AIDS
6. Byakika-Tusiime J, Oyugi J, Tumwikirize W, et al. Ability to purchase and secure stable therapy are significant predictors of non-adherence to antiretroviral therapy in Kampala, Uganda. Presented at: 10th Conference on Retroviruses and Opportunistic Infections; Boston, MA; 2003.
7. Orrell C, Bangsberg D, Badri M, et al. Adherence is not a barrier to delivery of HIV antiretroviral therapy in resource-poor countries. AIDS
8. Oyugi J, Byakika-Tusiime JB, Charlebois ED, et al. Multiple validated measures of adherence indicate high levels of adherence to generic HIV antiretroviral therapy in a resource-limited setting. J Acquir Immune Defic Syndr
9. Wenger N, Gifford A, Liu H, et al. Patient characteristics and attitudes associated with antiretroviral (AR) adherence. In: Program and Abstracts of the 6th Conference on Retroviruses and Opportunistic Infections (Chicago)
. Alexandria, Virginia: Foundation for Retrovirology and Human Health; 1999. Abstract 98.
10. Montessori V, Heath KV, Yip B, et al. Predictors of adherence with triple combination antiretroviral therapy [abstract 72]. Presented at: 7th Conference on Retroviruses and Opportunistic Infections; 30 January to February 2, 2000; San Francisco, CA.
11. Gifford AL, Bormann JE, Shivley MJ, et al. Predictors of self-reported adherence and plasma HIV concentrations in patients on multidrug antiretroviral regimens. J Acquir Immune Defic Syndr
12. Williams PL, Storm D, Montepiedra G, et al. Predictors of adherence to antiretroviral medications in children and adolescents with HIV infection. Pediatrics
13. Haynes RB. Determinants of compliance: the disease and the mechanics of treatment. In: Haynes RB, Sackett DL, eds. Compliance in Health Care
. Baltimore, MD: Johns Hopkins University Press; 1979:49-62.
14. Uzochukwu BSC, Onwujekwe OE, Onoka AC, et al. Determinants of non-adherence to subsidized anti-retroviral treatment in southeast Nigeria. Health Policy Plan
15. Horne R. Adherence to medication, a review of existing research. In: Myers L, Midence K, eds. Adherence to Treatment in Medical Conditions
. London, United Kingdom: Harwood Academic; 1998:285-310.
16. Yu JK, Chen SC, Wang KY, et al. True outcomes for patients on antiretroviral therapy who are ‘lost to follow-up’ in Malawi. Bull World Health Organ
17. Weidle PJ, Wamai N, Solberg P, et al. Adherence to antiretroviral therapy in a home-based AIDS care programme in rural Uganda. Lancet
© 2010 Lippincott Williams & Wilkins, Inc.