Adherence and barriers to HAART in Kiambu County, Kenya

Study design: We performed a cross-sectional study of HIV-positive patients from the Lusigetti Sub-County Hospital’s Comprehensive Care Clinic (LCCC) to assess the adherence rate and potential barriers to Highly Active Antiretroviral Therapy (HAART). Methods: Forty two percent of the clinic’s adult patient population (102 of 243) was surveyed over the course of three months using the validated Simplifi ed Medication Adherence Questionnaire (SMAQ), and ten additional questions (see appendix) to address barriers to adherence to HAART. Patient interviews were conducted in a private room, and responses to our survey were recorded without associated patient identifi ers. Participation was voluntary, no incentives were provided to patients, and oral informed consent was obtained. Results: Fifty three percent (53 of 100) of patients were identifi ed as adherent to HAART. Univariate logistic regression analysis revealed that the odds of adhering increased 2.8 fold with each level of greater perceived ease of access to medication refi lls (p= 0.0021). Conclusions: This study suggests patient adherence to HAART in Lusigetti Sub-county Hospital is much lower than expected based on previous patient viral load tests. Among the barriers assessed, poor access to medication was the only statistically signifi cant barrier to adherence. Since other countries in sub-Saharan Africa have similar processes for HIV medication refi ll and delivery, effective solutions for optimizing medication access could be implemented across other African nations in order to increase HAART adherence and improve the control of HIV. Research Article Adherence and barriers to HAART in Kiambu County, Kenya Malek Bashti1, Eduardo Palacios Fabre2 and Lars Osterberg3* 1Life Science Research Professional, Stanford University School of Medicine, California, USA 2Epidemiology and Clinical Research Professional, Stanford University School of Medicine, California, USA 3Associate Professor (Teaching) of Medicine, Stanford University School of Medicine, California, USA Received: 13 May, 2021 Accepted: 02 June, 2021 Published: 03 June, 2021 *Corresponding author: Lars Osterberg, MD, MPH, Associate Professor (Teaching) of Medicine, Stanford University School of Medicine, California, 1265 Welch Rd. X152, Stanford, CA 94305, USA, Tel: 650-723-1536; E-mail:


Introduction
In Kenya, 1.6 million adults (ages 15-49) are living with HIV (a prevalence of 4.7% of the adult population) [1]. However, this prevalence varies signifi cantly between Kenya's 47 counties with Wajir County at 0.1% to Siaya County at 21.0% of the population, and globally Kenya has the twelfth highest adult prevalence of HIV [2]. Nonetheless, in the past few decades, Kenya has been a success story in advancing HIV prevention efforts as one of the fi rst to approve the use of pre-exposure prophylaxis and has led the way in providing voluntary medical male circumcision. AIDS-related deaths dropped from 53,900 in 2010 to 28,200 in 2017. This change is directly attributable to the government's rollout of free Highly Active Antiretroviral Therapy (HAART) starting in 2003, and the adoption of the World Health Organization's (WHO) recommendations to immediately offer treatment to people diagnosed with HIV [3].
In order to effectively suppress viral load and prevent both the transmission of disease and the emergence of drug resistant strains, patients with HIV require adherence rates of 95% or greater, and studies of medication adherence to HAART in Kenya have shown adherence rates range between 43-90% [4][5][6].
Understanding the barriers to and predictors of HAART adherence is an important public health goal in helping to control HIV, particularly now that the WHO recommends HAART for all persons regardless of CD4 counts [4]

Ethical considerations
This study was conducted after obtaining offi cial approval from the Kiambu County Governor's Offi ce and the director of Lusigetti Sub-County Hospital. Participation was voluntary, no incentives were provided to patients, and oral informed consent was obtained from all who participated in the study; no patient identifi ers were collected to protect patient privacy. This project was approved by the Stanford Institutional Review Board (Human Subjects).

Study design
Study patients were identifi ed when they presented to LCCC for their scheduled refi ll appointment and invited to participate in this research project. Patients were excluded from participating in the study if they had been prescribed HAART fewer than three months prior to consent or if they were under 18 years of age. Informed consent was obtained through a scripted verbal informed consent process, which was also translated to Kiswahili. All patients who were invited and eligible, agreed to consent to the study and completed the survey, giving a response rate of 100%. To ensure patient anonymity and confi dentiality, patient interviews were conducted in a private room, and responses to our survey were recorded without associated patient identifi ers. The survey included the validated Simplifi ed Medication Adherence Questionnaire (SMAQ), and ten additional questions (see appendix) to specifi cally address barriers to adherence [7]. The SMAQ classifi ed patients as non-adherent when they answered any of the qualitative questions in a non-adherent manner, missed more than 2 doses over the past week, or over 2 days of total non-medication during the past 3 months [7].

Results
One hundred two patients were interviewed over the course of ten weeks, representing 41% of the clinic's patient population. Two patients did not meet inclusion criteria as they were prescribed HAART less than 3 months prior to the interview and were not included in the study. According to the SMAQ component of the survey, 53% (n=53) of patients were adherent to their medication regimen. The questions within the SMAQ that frequently yielded non-adherent answers were, "Have you ever forgotten to take your medication?" (n=27) and, "When you feel bad, have you ever discontinued taking your medication?" (n=20).
Patients were asked to rank the ease of refi lling their medication on a Likert scale: very diffi cult, diffi cult, easy, and very easy. Univariate logistic regression analysis revealed that the odds of adherence increased 2.8 fold with each level of greater perceived ease of access to medication refi lls (p= 0.0021).
In order to assess patients' perspectives on the importance of taking their medication regularly they were asked to select one of the following: not at all, indifferent, somewhat, and very. Furthermore, patients were asked to qualify how they think their health has changed since starting HAART: gotten worse, stayed the same, somewhat improved, improved a lot. Patients who believed taking their medication regularly was important presented 1.7 greater odds of believing that their health had changed for the better since starting HAART, but this was not statistically signifi cant (p = 0.0688) ( Table 1 for summary of results).
Comparison of mean time in years for the duration of being HIV-positive, time since fi rst prescribed HAART, and the time difference between diagnosis and commencement of treatment between adherent and non-adherent patients.
Summary of the effects of barriers to adherence on the odds ratio that patients remain adherent to HAART.

Discussion
This study sought to determine the adherence rate and barriers to adherence in patients on HAART attending the LCCC. The high response rate of 100% in this study is in part likely due to the cultural excitement of interacting with a "mzungu," meaning white or foreign person, often used affectionately and can contribute to patients' eagerness to participate in a study by foreigners. In addition, the face to face nature of the interview and having the research study separated from the care patients received at LCCC were also essential in establishing trust and contributing to the high response rate.
Prior to this survey, the clinic staff believed 94% of their patients were adhering to their medication regimen, based on extrapolation from viral load tests. It is not uncommon for patients to improve their medication taking behavior around appointment time, a behavior described in the literature as white coat adherence where patients have improved medication adherence around the period of clinic visits and blood testing [8]. In anticipation of upcoming blood tests, patients prescribed HAART can achieve a viral load <400 copies/mL in under four weeks and give the impression they are adherent to their medications when they may not be taking their medications as prescribed at times outside of the 4 week clinic appointment times [9]. The healthcare teams at LCCC occasionally do pill counts to assess adherence, but this is also an unreliable means of measuring adherence as some patients are known to "pill dump" prior to appointments in an effort to appear more adherent to their medications. Sub-optimal adherence is unhealthy for patients as this can lead to treatment failure, poor survival outcomes, and drug resistance. There is also a threat to the community as high viral loads increase the risk of both HIV transmission to others and increases in HAART resistance [10].
Our results may be biased given patients were recruited at their clinic appointments and patients who adhere to clinic appointments are also more likely to adhere to their medications. Our measured adherence rates may therefore overestimate the true adherence rates in the general clinic population. One reason for the poorer adherence rates found in our study compared to previously reported adherence rates in Kenya might be that patients in our study were more willing to be forthcoming about missing doses due to the reasons mentioned above, and due to the fact that patients could remain anonymous and not identifi ed by clinic staff. The SMAQ is also a sensitive instrument in identifying non-adherent patients given one positive response to the questionnaire classifi es a patient as non-adherent. Our results are not much different to other region-specifi c studies of adherence in Kenya and from UNAIDS data from 2017 suggesting that only 51% of HIVpositive people living in Kenya have complete suppression of viral load. [6] This data is important for health care providers across Kenya as they consider including measures of adherence implemented in between clinic visits. Other options include using more objective measures such as Medication Event Monitoring System (MEMS) devices or using multiple modalities of adherence measures to more fully understand the medication taking behavior of patients.
Access to medication was the only statistically signifi cant barrier to adherence, and gaps in care provision is a wellknown factor impacting medication adherence. Anecdotal patient reports revealed that LCCC patients would prefer the option of refi lling their medication in larger batches and less frequently to minimize their time off work to travel to the clinic; future research should focus on fi nding the other factors that make the refi ll process diffi cult for patients. This method of Multi-Month Scripting (MMS) is already in place at LCCC, however, more research needs to be done to improve the implementation of MMS and other processes to optimize delivery of medications to patients. Since other countries in sub-Saharan Africa have similar processes for the refi ll and delivery of HIV medications, effective solutions discovered through further research endeavors could be implemented across other African nations. We must empower patients with knowledge and resources for optimal medication management and provide them with the support they need to achieve the full benefi t of these highly effective medications.