Recruitment, sampling, and data collection.

The study population was a cohort of male and female adolescents who were 10 to 19 years of age. We enrolled adolescents that were on ART, on IPT (patient must have started according to national guidelines), known age at the start of IPT initiation, had at least 2 medication refill appointments within the pre-and post-implementation periods and had continued with IPT for 6 months following initiation [28, 29]. We excluded adolescents that were initiated on IPT but were lost to follow-up or discontinued IPT—either self-directed or as per the medical provider’s guidance.

We made certain that participants’ HIV status was fully disclosed during the qualitative component. Disclosure status was verbally verified before the in-depth interviews by asking an adolescent why they take medications daily. Furthermore, we used extreme deviant case sampling by including participants that reported either extremely good adherence (pill count between 95%-100%), or extremely poor adherence (pill count less than 75% or above 125%), and finally, adolescents who were also included in the quantitative cohort and either (1) attend Teen Club or (2) still follow at BCMCFM-COE. The study only included participants on the first-line and second-line ART regimens. This was the case because first and second-line ART regimens are relatively easy to take compared to third-line or non-standard regimens.

Healthcare workers who have worked with ALHIV for at least five years were included as study participants. To improve the variability of Healthcare worker responses, clinicians, nurses, and psychosocial workers were offered the opportunity to take part in the study.

For the quantitative component of the study, we used the following formula n = Z² P(1-p) / d² [33, 34] to calculate sample size (n = sample size, p = expected proportion of non-adherent clients, Z = confidence interval—statistic corresponding to the level of confidence, d = margin of error—corresponding to effect size). Using national adherence rates in adolescents (45%), a total of 385 medical records that met the inclusion criteria were selected for review using a computer-generated simple random selection technique to represent active clients at the facility [4]. For the qualitative component, we followed a purposive sampling technique to select 16 ALHIV and three healthcare providers for in-depth interviews and key informant interviews respectively. The qualitative sample size was based on assertions from Guest et al. who stated that by the 12^th interview one would have reached saturation. However, in our study, we included 16 participants because of the distribution of the variables of interest in our purposive sampling approach [35]. ALHIV were selected according to their adherence pattern which was determined by physical pill counting or reported pill count. Eight participants that were regarded as very adherent were selected together with eight participants who usually reported suboptimal adherence. Adolescents who met the criteria and were in care at the study site between the period of 1 February 2017 and 31 January 2018 were considered for study participation. Healthcare providers were also purposively selected based on their involvement in the provision of care to the adolescents at the study facility. The non-probability sampling method was deliberately selected to have a sample that contains characters of interest [36].

Quantitative data comprised a retrospective review of prospectively collected data of ALHIV who were in care between 1 February 2017 and 31 January 2018. A checklist containing variables of interest; age, sex, ART regimen, IPT, and Cotrimoxazole preventive therapy (CPT) status, disclosure status, education level, and religion was used to extract medical records from Electronic Medical Records (EMR). The study site uses EMRx which has a variety of features for the clinic, laboratory, psychosocial, and pharmacy, among others, to record medical records at each hospital visit. Some of the participants’ records selected for review in pre IPT period were not included in the post IPT review if they did not satisfy the inclusion criteria.

Qualitative data was collected through in-depth interviews and key informant interviews using semi-structured interview guides. Participants were approached during teen clubs where they were asked to participate in the study after accessing all the services they came to the clinic for. Healthcare providers were interviewed as key informants as they have been involved in the care of ALHIV and have a vast knowledge of adolescent behaviors towards medication. We deliberately interviewed the participants after accessing services and health care workers at their convenient time to realize undivided attention during interviews and to avoid disruption of work respectively. Once a participant signed a consent form a semi-structured interview guide was used to ask questions. The interviews were audio-recorded to capture every detail.

The interview guides were in both Chichewa and English and were utilized according to the language the participant understands and preferred. The interview guides included questions about medication knowledge, facilitators and barriers to adherence to ART, and general perceptions of IPT’s inclusion in the HIV care package. The interview guide for healthcare workers further included questions about the general training and healthcare knowledge on IPT and its side effects. Interviews were conducted in a secure, closed room, and participants were identified by assigned codes as a measure of safeguarding their privacy.

To minimize the threats to the credibility of the study, the investigator used member checks and triangulation approaches [37, 38]. Credibility refers to the confidence that can be placed in the truth of the research findings by establishing whether research findings represent plausible information drawn from the participants’ original data [39, 40]. During the interviews, the information given by the participants was summarized before asking them to verify if the captured information was accurate. Member-checking ensured that the information collected is shared with participants so that intentions are clarified and errors if any, are corrected. We also mixed the data from the interviews with the secondary data to enhance cross-data validity checks [37]. Furthermore, the PI who is an experienced HIV clinician also engaged and trained an additional experienced clinician to help with face-to-face interviews while maintaining strict confidentiality throughout the process. Researchers who were involved in data collection took notes about participants’ comments and researchers’ thoughts during interviews. Additionally, reflective notes from the data were recorded soon after the interviews (memoing). This practice was vital to the process of bracketing and reflexivity in mitigating the potentially deleterious effects of unacknowledged preconceptions related to the research and thereby increasing the rigor of the research [41, 42]. The PI verified all the transcripts that were translated from Chichewa to English for completeness and relevance. We also triangulated the analysis approach by using multiple data analysts. A second analyst who was not involved in data collection independently reviewed the interview data. Any gaps or blind spots were discussed with the PI, and the aggregated results were documented. Using multiple investigators throughout data collection and analysis reduced the risk of bias caused by a single person doing all the tasks.

Quantitative data.

Quantitative data that was extracted from EMR was cleaned using an excel package. To ensure the confidentiality of the participants, we assigned unique numbers to each participants’ record. Both data sets were inspected for completeness before actual analysis. The data in Excel was then exported to R Studio (Ver. 1.3.959) for analysis.

The primary outcome variable for this study was ART adherence and it was measured as a dichotomous variable. All continuous variables such as the age of participants were analyzed using frequency distribution graphs. Descriptive statistical methods such as means, standard deviation, and proportion were generated to further examine our population. Furthermore, statistical analyses using 2-group and regression analysis were executed to measure the association of pill burden, gender, ART regimen, HIV disclosure status, and ART adherence. Variables of interest namely; ART regimen, gender, disclosure status, and IPT status were included in the final multivariate analysis using a forward stepwise approach, regardless of whether they were statistically significant or not in the univariate analyses. The factors were included in the logistic regression model because theoretically these variables are known to interact with each other in adolescent health which can impact adherence [1, 2]. A two-group test using χ² was executed to make crude comparisons of ART adherence before and after IPT introduction and its associated factors with a p-value set at 0.05. Logistic regression analysis was used to quantify the association between IPT inclusion and ART adherence.

ART adherence was defined as taking at least 95% of the doses (and no more than 105% from losses) at the prescribed interval and within the refill period to prevent HIV drug resistance [28]. In situations where ART pills were not brought to the clinic for pill-counting, missing no more than 3 doses in a month was also defined as good adherence. We favored logistic regression in this study because our outcome of interest was dichotomous. Furthermore, it allowed for statistical control of known confounders such as concurrent adherence reinforcement interventions; this is a particularly important factor because we used a before and after study design.

Qualitative data.

We employed a thematic analysis of the data collected through in-depth interviews and key informant interviews as suggested by Braun and Clarke [43]. All audio recordings were translated and transcribed verbatim simultaneously if the interview was not done in English. All interviews conducted in English were immediately transcribed. Chichewa audios were transcribed straight into English by a transcriber conversant in English and Chichewa and were verified by the PI against the audios to ascertain quality and correctness. The researchers manually analyzed the data. The researchers initiated the process of immersion and familiarization with the data while conducting the interviews, and later on, read and re-read the transcripts multiple times to understand the depth of the content. We inductively and deductively coded the data. The inductive codes were drawn from the collected data while theoretical codes were deduced from the interview guide and study objectives. We identified abstract themes from the coded text segments that formed a similar coherent familiar pattern, which was further organized to develop themes. Then the researchers reviewed and refined potential themes to match the rest of the data, where the themes were not complete or did not match the rest of the data or objectives, additional coding was done. The thematic analysis allowed the researchers to interpret and summarize the themes for a report to be produced. Both data types were utilized in a narrative integration manner where we used a weaving approach to present data on a concept by concept basis [44].

Characteristics of the quantitative sample.

The social demographic characteristics of the participants based on the data extracted from EMR from 1 February 2017 to 31 January 2018 are summarized in Table 1 below. There were a total of 385 participants, mean age of 14.3 years (SD = 2.7). One hundred and ninety-three participants had their records analyzed before the introduction of IPT while 192 had records analyzed after the introduction of IPT into the ART package. Ninety-one percent of participants (n = 349) were on first-line ART at the time of the study, while the rest were on second-line ART.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of sample demographics.

https://doi.org/10.1371/journal.pgph.0000418.t001

Demographic characteristics of study participants.

We interviewed 16 adolescents aged between 10 and 18 years who are living with HIV and were in care at BCMCFM-COE and three health workers. The median age of the adolescent was 14 with an interquartile range of five (12–17). Out of the 16 adolescents interviewed, eight were females. Eight of the adolescents reported good ART adherence at least 3 months before the interview and the rest reported suboptimal adherences. All adolescents attended school but only five were in secondary school. Only six adolescents were living with non-biological caregivers and all adolescents belonged to religious affiliation, five of whom were Muslims. All adolescents were on IPT at the time of the interview and 13 had full disclosure on their HIV status-completed as per guidelines while three had completed a partial disclosure of their HIV status. Partial disclosure refers to status disclosure in which the client becomes aware of an existing health problem without necessarily knowing the actual condition. It acts as a preparatory step in HIV status disclosure for children. Table 2 summarises the demographic characteristics of participants who were interviewed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographic profile of participants who were interviewed.

https://doi.org/10.1371/journal.pgph.0000418.t002

Of the three health care workers interviewed (Key informant interviews), one was a senior clinician and ART provider, one was an adolescent nurse and the last one was a social worker. All the health workers had worked with adolescents for over five years and were directly involved in the care of adolescents both in the clinic and teen club activities. The social worker was particularly involved in developing curriculums for teen club lessons on life skills and psychosocial issues. She was also responsible for handling any psychosocial issues affecting adolescents on ART, including planning and carrying out disclosure and adherence counseling sessions for adolescents.

Factors that affect adherence.

In general, the non-adherence rate to ART was 27.5% in the pre IPT period, and 32.3% after IPT was added to the ART package. On factors associated with ART adherence before IPT, only ART regimen (p-value = 0.0462) was associated with ART adherence leaving out gender, (p-value = 0.10) and disclosure status (p-value = 0.55). ART regimen (p-value = 0.0537) was also associated with adherence after IPT was introduced (Tables 3 and 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Bivariate analysis of factors associated with ART adherence before and after added pill burden (IPT introduction).

https://doi.org/10.1371/journal.pgph.0000418.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Multivariate logistic regression model of factors associated with ART adherence.

https://doi.org/10.1371/journal.pgph.0000418.t004

Multivariate analysis of factors associated with adherence in adolescents showed that only the ART regimen was somewhat associated with adherence. OR 2.11, 95% CI 0.97–4.53, p-value = 0.0537. (see Table 4). However, since the 95% confidence interval includes a null value, the difference in ART regimen doesn’t affect ART adherence. Consequently, the ART regimen was not associated with ART adherence in adolescents living with HIV/AIDS at Baylor Children’s Centre of Excellence.

Adherence before and after IPT introduction (Added pill burden).

Overall, females were more adherent to ART than males at 72% (149/207) versus 68% (121/178) in the period of study (6 months before, and 6months after IPT introduction). However, the difference in adherence between males and females was not statistically significant (χ² = 0.7321, p-value = 0.39). Seventy-one percent (249/349) of clients on the first-line regimen reported good adherence while 58% (21/36) of clients on the second-line had acceptable adherence. Although ART adherence was different as per the ART regimen, the overall difference was not statistically significant (p-value = 0.10). After IPT was introduced, female participants maintained optimal adherence 53.1% (69/130), than males 46.9% (61/130). However, the difference in ART adherence between males and females after added pill burden was not statistically significant (p-value = 0.392).

Pill burden. Participants stated the addition of preventive therapies within HIV management such as Isoniazid for prevention of TB and Cotrimoxazole for prevention of Pneumonia and some diarrheal diseases significantly increased the number of tablets. The majority of participants stated that they regarded IPT as a drug outside the ART package and considered it as not important. Therefore, IPT added an unnecessary pill burden. The participants narrated that:

Yea, IPT somehow affected the number of pills to take in a day, but not necessarily on ART, because most of the time I would take ART and skip IPT. Male 4

Sometimes, I could feel I had many tablets to take but it was still manageable because I had about 4 tablets to take at a go. However, I got used to the number of tablets so everything was okay. Male 2

Health workers corroborated the adolescents’ view and stated that the introduction of IPT led to adherence problems where adolescents could skip all medications, or at times only adhere to ART and Cotrimoxazole, but skip IPT. This was the case because of the added pill burden coupled with poor client preparation on anticipated side effects and increased number of tablets as explained in the following excerpts:

There wasn’t much of a change in ART adherence, but we realized that most of them were concentrating on ART and CPT and not IPT. It looks like when they thought IPT was bothering them due to the pill burden, they only concentrated on the old meds. HCW III

The burden from the additional pills was even worse for adolescents on complex ART regimens as stated by one health worker:

I feel they did not have the urge to take IPT. It is the case because I feel IPT was implemented in a hurry. We didn’t have enough time to prepare them. After all, when we introduced it we used to tell them that it helps prevent TB in those whose immunity is low. Most of our adolescents are on second-line ART with complex regimens. HCW III

ART regimen and adherence. Overall, participants on first-line ART had better adherence (92.2%, 249/349) compared to those on second-line ART (7.8%, 21/36, p-value = 0.104). This trend was also true even when the participants were split into before and after IPT cohorts. There was a significant difference in adherence between first and second-line ART regimens in females after IPT was introduced (72.6%, 69/95, p-value = 0.0462).

Disclosure status and ART adherence. Eighty-three percent (223/314) of fully disclosed participants showed optimal adherence compared to their non-disclosed counterparts. This difference was not statistically significant when the variables were analyzed as either combined or split cohorts. (p-value = 0.55). The bivariate analysis did not show any effect of HIV disclosure status on overall ART adherence.

Lack of HIV status disclosure. Participants who live with relatives or friends to whom they have not disclosed their HIV status expressed difficulties in keeping up with ART adherence. This is because, unlike other medications, ARVs have a stigma attached to them. As a result, ALHIV willfully forego medication when in the company of undisclosed acquaintances or relatives. The following excerpts indicate this finding:

Taking my medications is problematic when I am busy, sometimes when I am among friends it is hard to take my medication too. I don’t want them to ask many questions and know about my status. Female 7

HIV status disclosure remains a challenge even among close family members. To save marriages, some guardians prefer to conceal their HIV status or that of their children. One participant stated that her mom categorically indicated that she should never mention her HIV status when her stepdad is around which prompts her to willfully skip her doses whenever her stepdad visits them. This is what she said:

Having visitors at home or when my stepdad comes home, it becomes an issue to take my medications. He [Stepdad] doesn’t know that I am on ARVs and my mom said we should keep it that way. Female 3

Other factors that influenced adherence that were only apparent in the qualitative analysis were as follows:

Pressure secondary to academic responsibilities. All the respondents highlighted school activities as one of the barriers to good adherence. Many issues emanated from those participants in boarding schools who stated that keeping up with activities such as daily study periods brought a lot of problems with adherence. For instance, adolescents stated that their evening study periods in school collided with evening ART doses to the extent that most participants could end up missing the doses. Further to this, the quest to take ARVs without schoolmates noticing was a major challenge for most participants. These findings are explained in the following excerpts:

When I have school work to do and a tight schedule I forget my medication. This is usually the case because I have to get up early, prepare for school and I end up forgetting to take my pills. Female 4

One health worker observed that most suboptimal adherence rates are recorded during school sessions for those in boarding schools and at the level of taking national examinations. It was further stated that the lack of monitoring among boarders that are on ART contributes to suboptimal adherence.

School activities interfere with adherence greatly. Even older teens do complain that studies and other activities are quite hard to accommodate adherence. Even going on school holiday has been a very big issue as there is no or little monitoring. Parental support and continued teen club sessions have always been key to good adherence in our school-going teens. HCW I

Peer pressure. One health worker affirmed that lack of disclosure is indeed a barrier to achieving optimal ART adherence, and it extends to partners. He further observed that peer pressure among adolescents is an important factor that has the potential to hinder good adherence. Adolescents who engage in love relationships because of pressure from friends are likely to hide their HIV status from partners. This further means sometimes skipping doses when they are close to their lovers for fear of being questioned which has the potential of leading to unintended disclosure.

…love relationships among adolescents worsen adherence because partners do not usually disclose to each other for fear of losing love partners. Peer pressure from friends is also an important barrier. When they are interacting some non-adherent peers give tips to friends on pill counting so that they show good adherence at the clinic when they are skipping doses. HCW III

Poor support. Adolescents mentioned a lack of parental, peer, or sibling support as a barrier to adherence. Adolescents who stayed with non-supportive parents and peers reported difficulties adhering to medication. This is what they said:

When I fight with my siblings I get so angry to the extent that I don’t take my meds… “panthawi imeneyo ndimakhala kuti mankhwala ndawataya uko!” (At that moment, I don’t want anything to do with my medication). Also when I disagree with my mom it’s the same, I don’t take my meds. But these have nothing to do with IPT. Male 2

When I fight with my siblings they throw my drugs out of the house. They are mean to me sometimes, so that affects me. Otherwise, the number of pills never bothers me. Male 1

Health workers stated that non-adherent peers give tips to adherent ones such as pill dumping and pill calculations to show good adherence at the clinic when they are not taking their medications.

Peer pressure from friends is a negative influence on adolescents. When they are interacting some non-adherent adolescents give tips to show good adherence at the clinic. HCW III

Patient-level factors.

Self-reinforcement techniques. Participants stated that they fail to take medication due to forgetfulness. To deal with forgetfulness, participants employed several measures for them to remember when to take medicines including the setting of alarms. Phone alarms are commonly used because phones are more widely available than proper clock alarms. Another tactic used by adolescents to remember to take their medicines was the act of keeping medications where one can easily access them. Some participants reported that they kept their medication bottles in their pillowcases or shoes. Participants explained that it is easy to remember to take medication at night once they hear the clanking sound of the ART bottles in the pillowcase. The explanation was similar for those who keep medication in the shoes to remember in the morning before going to school as indicated in the excerpts below:

I use a calendar and tick it every time I take my meds, when I am at school I sleep with my pills so that I should remember in the morning. Male 4

I use the alarm on my mom’s phone, sometimes when the phone is off, I use a wall clock. I also put my medication in my school shoes when all these things are not available so that I take them in the morning before going to school because I keep my medication. Female 6

The longer duration of ART. Participants stated that being on ARVs for longer duration makes one more responsible as one becomes fully aware of the need for ARVs. They explained that unlike those who are just starting ART and are heavily reliant on parents, adolescents on ART for a longer duration must have gone through ART and adherence sessions more than once such that they are familiar with adherence issues.

Being reminded by my parents but also just being responsible. I have been on medication for years and I know I have to take my medications every day. So I make sure I take my ARVs after all I keep my pills myself. Female 6

Health system and social factors.

The health system and social factors that facilitate adherence to ART among adolescents include Teen Club services, and peer, guardian, and clinic-based psychosocial support.

Teen club activities. It overtly surfaced that teen clubs played a great role in reinforcing adherence among adolescents. The lessons that are delivered at the teen club, peer experiences, the sense of belonging, and the feeling that there are equally other teens on ART who are doing well help adolescents to cope with their HIV status and be adherent to medications. One participant stated that despite the different adherence techniques he uses; the teen club has been instrumental in maintaining good adherence because of the adherence lessons offered. Many participants further explained that the lessons help them not only to remember to take medication but also to take good care of themselves both at home and school. This is what participants said:

With exposure to a teen club, things are different. My adherence used to be poor because I was not exposed to life skills coaches and I was not feeling encouraged. It’s different from the time before I joined the teen club. Female 8

Further to several techniques we use, I think a teen club is one important thing holding us adolescents. Because when we come here we end up learning one or two things about how we can look after ourselves, how to take medications, and deal with other important issues such as social stigma at school. When you are here you learn more and make important and good decisions. Male 2

Peer, guardian, and clinic-based support. Parental/guardian and peer support also emerged as an important facilitator to medication adherence. The majority of participants stated that parental or guardian support was vital to adherence. Participants stated that being reminded of the time to take medication and getting support such as transport money to and from the clinic on refill appointment dates were perceived to be important. One participant observed that being reminded and being shouted at by his guardian whenever he misses doses helps him to be more adherent to his medication. This was further expounded by one participant who stated that he found comfort in the fact that he was not alone taking ARVs in the house because his two siblings were also on ARVs which encouraged him to be adherent to his ARVs.

The following were some of the responses:

Having relatives who are also taking drugs is very important. They remind you. Family support through reminders and keeping drugs in places where you can easily see them helps to remind you that you need to take meds. My parents remind me to take my meds when they are taking their meds. Female 3

Being reminded by my mom and being shouted at is a constant motivation to me. I know that if I skip my medication my mom will shout at me and that alone reminds me to take my medication all the time. Female 5

Health workers affirmed the role of psychosocial counseling, peer interaction during teen clubs as well as building rapport with adolescents as key factors that facilitate good adherence. One healthcare worker narrated:

Well for me, interaction with adolescents is important. They can take our advice because we are in a good relationship with them, Secondly, the fact that we temporarily suspend them from attending teen club when they are non-adherent also helps them to keep up the good work so that they don’t miss Teen Club. And the frequency of refill visits also helps them to be adherent. Those who are non-adherent have to come to the clinic frequently. HCW III