Promoting Tuberculosis Preventive Therapy for People Living with HIV in South Africa: Interventions Hindered by Complicated Clinical Guidelines and Imbalanced Patient-Provider Dynamics

Abstract

Isoniazid preventive therapy (IPT) reduces the risk of active tuberculosis among people living with HIV, but implementation of IPT in South Africa and elsewhere remains slow. The objective of this study was to examine both nurse perceptions of clinical mentorship and patient perceptions of in-queue health education for promoting IPT uptake in Potchefstroom, South Africa. We measured adoption, fidelity, acceptability, and sustainability of the interventions using both quantitative and qualitative methods. Adoption, fidelity, and acceptability of the interventions were moderately high. However, nurses believed they could not sustain their increased prescriptions of IPT, and though many patients intended to ask nurses about IPT, few did. Most patients attributed their behavior to an imbalance of patient-provider power. National IPT guidelines should be unambiguous and easily implemented after minimal training on patient eligibility and appropriate medication durations, nurse-patient dynamics should empower the patient, and district-level support and monitoring should be implemented.

Introduction

Over 7.1 million people in South Africa were living with human immunodeficiency virus (HIV) in 2016, more than any other country worldwide [1]. Among people living with HIV, the leading cause of mortality is tuberculosis (TB) [1, 2]. The World Health Organization (WHO) recommends isoniazid preventative therapy (IPT) to prevent TB in people living with HIV [3].

IPT is effective and can reduce the risk of active TB by as much as 60% among people living with HIV who have evidence of a latent TB infection [4]. The WHO currently recommends that all HIV-positive individuals without active TB in resource-constrained, high HIV- and TB-prevalence settings should be treated with 36 months of IPT [3]. In 2015, the Department of Health in South Africa rolled out national IPT guidelines based on the WHO recommendations but diverged in two ways [5]. First, a tuberculin skin test (TST) was required within one month of initiating IPT and second, patient use of antiretroviral therapy (ART) determined the prescribed duration of IPT. Despite the WHO’s recommendations and South Africa’s initiative to provide IPT, implementation has been slow. Less than 51% of patients newly enrolled in HIV care started IPT in 2016, and in South Africa, these numbers may be overstated [2, 6].

Since the release of the 2015 IPT guidelines in South Africa, many barriers such as including poor patient-provider communication and insufficient clinic staffing have constrained uptake [7–10]. Per previous qualitative work in in the Gauteng Province of South Africa, the most substantial obstacle for clinicians in South Africa to prescribing IPT is a lack of confidence in their ability to write appropriate prescriptions and the treatment [7]. Interviews revealed that clinicians did not trust IPT to effectively prevent active TB. Instead, clinicians believed that IPT would be prescribed to those with active TB unintentionally and hence promote drug resistant TB. Even among clinicians who prescribed IPT to their patients, implementation was inconsistent, with some restricting their prescriptions to patients on ART or with CD4 counts below 200 cells per milliliter. Further compounding the issue, patients are rarely familiar with IPT or its benefits [7]. These barriers suggest that efforts to expand the uptake of IPT will continue to achieve limited outcomes without acceptable and feasible interventions that motivate clinicians and patients to use this critically important treatment.

In this study, we piloted a clinical mentorship intervention to improve nurse willingness to prescribe IPT and an in-queue health education intervention to improve patient awareness of and empowerment to request IPT. We chose to intervene on both nurses and patients to generate supply and demand of IPT respectively [11]. The objectives were to measure adoption and fidelity of the intervention, to assess nurse and patient perceptions of the acceptability and sustainability of the intervention, and to examine continued barriers to change.

Methods

Study setting and population

We conducted this study in peri-urban areas surrounding the city of Potchefstroom in South Africa. Potchefstroom is in the Dr. Kenneth Kaunda district, where HIV prevalence is 12.9% and TB incidence is 696 per 100,000 population [12, 13]. Unemployment in the district is 29.7%, and a minority (10%) of residents have had no formal education [14]. The district is located in the City of Matlosana Municipality, where the majority of residents speak multiple languages with the most common primary languages being Setswana (36%), Sesotho (20%), Afrikaans (17.3%), Xhosa (14%), and English (4.5%) [15].

We chose to implement our intervention in the four largest of nine public clinics — as measured by number of patients on ART in 2015 — in which there were no other ongoing IPT-related interventions. The four study clinics each had 500–600 patients on ART monthly. Three clinics were larger, with 1,000 patients on average, but had concurrent IPT-related interventions occurring and were excluded from the study. Using a random number generator, we evenly assigned clinics to receive one of the following: a nurse-centered or patient-centered intervention.

We conducted a multi-method assessment of the interventions. The quantitative assessment was conducted at a clinic level. For the qualitative assessment, we recruited all eligible, consenting nurses at clinics assigned to the nurse-centered intervention and a convenience sample of eligible patients from clinics assigned to the patient-centered intervention. Nurses were eligible to participate if they had provided HIV care at their clinic for the past two months, were able to prescribe ART, and had participated in the nurse-centered intervention. Patients were eligible to participate if they were ≥ 18 years, HIV-positive, had received care at the intervention clinic for at least two months, and had participated in the patient-centered intervention. All research participants provided written informed consent and received 50 Rand (approximately $3.50 USD) for their time. This study was approved by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board and the University of the Witwatersrand Human Research Ethics Committee.

Description of the Interventions

Nurse-Centered Intervention

The nurse-centered intervention promoted two target behaviors: checking IPT-eligibility for all HIV-positive patients and prescribing IPT to all eligible patients. A clinical mentor, who was a certified professional nurse with six years of previous clinical experience, led the nurse-centered intervention. The intervention consisted of six components: (1) a workshop to increase knowledge about IPT and to list clinic-specific barriers to prescribing IPT, (2) in-room consultations, (3) a workshop to select key barriers and to create an Action Plan for those barriers, (4) aid in electing and training an IPT Champion, (5) communication with the district-level TB/HIV coordinator, and (6) remote consultations via text messaging (Figure 1). Details and theoretical groundings for each component are presented in Table I.

Figure 1:

Intervention Timeline

Table I:

Components of the Nurse-Centered Intervention Led by the Clinical Mentor

Component	Week of Study	Frequency (Average duration)	Total Hours (% of Hours)	Description	Theoretical Basis
Leading Group Workshop #1: Education and Reflection	1–2	Once per clinic (90 minutes), including 1 hour and 15 minutes more for absent nurses	4.3 (4.5)	Led exercise to map individual procedures for HIV-patients; teach and role play national guidelines; brainstorm clinic-specific barriers	—
Mentoring	1–14	36 sessions with 11 nurses (88 m)	55.8 (58.8)	Modeled behavior to individual nurses and support patient interactions	Modeling target behaviors to promote self-efficacy through social learning, from Bandura’s social cognitive theory [17], [28]. WHO Guidelines for Clinical Mentoring for HIV Support in Resource Constrained Settings [10]
Leading Group Workshop #2: Action Plan	3	Once per clinic (55m), including 1h 15m more for absent nurses	3.1 (3.3)	Brainstormed solutions to clinic-specific barriers and selected a few solutions to create an Action Plan	Results-based management, from the United Nations Development Group [31] Theory of Change, from the Aspen Institute [32]
Training the IPT Champion	4	3 times (31.7 m)	1.6 (1.7)	Elected a nurse from the clinic, i.e. an “IPT Champion,” to continue the Action Plan after the intervention ended	The Expert Recommendations for Implementing Change (ERIC) Project [30]
Assisting with the Action Plan	4–14	21 times (71.2 m)	24.9 (26.3)	Worked with operations manager, pharmacist, data capturer, and clerk to implement Action Plans	—
Communicating with the district-level coordinator	8	Once (72 m)	1.3 (1.3)	Explained barriers to subdistrict HIV/TB coordinator and requested additional support	—
Supporting nurses via text messaging	18–20	6 text messages sent; messages read within 19 minutes of receipt on average	4.0 (4.2)	Sent text messages	Cues to Action, from the Health Behavior Model [33]

In brief, the clinical mentor taught and reinforced the South African IPT guidelines in the first workshop through roleplay and brainstormed clinic-specific barriers that prevented the target behaviors (Supplemental Material). In the weeks following the first workshop, the clinical mentor provided in-room consultations during the provision of clinical care to patients to model the target behaviors and to better understand clinic-specific barriers to IPT implementation. The clinical mentor then led a second workshop where nurses created an Action Plan with solutions to address clinic-specific barriers. Nurses then nominated and elected a fellow nurse to be an IPT Champion in each clinic. The IPT champion was responsible for implementing the Action Plan with help from the clinical mentor and continuing the Action Plan after the end of the intervention. For the last three weeks of the study, the clinical mentor stopped providing in-person consultations and instead contacted nurses by mobile phone through a text-messaging platform. Messages acted as a reminder to perform the two target behaviors, and offered continued, two-way communication opportunities for sustained educational and clinical support.

Patient-Centered Intervention

The patient-centered intervention promoted one target behavior: patients should ask a nurse about IPT. The patient intervention consisted of three components: (1) in-person, oral health education sessions for patients waiting in the queue, (2) the provision of calendars printed with relevant health education information to incentivize participation in those education sessions, and (3) posters in each clinic prompting HIV-positive patients to ask their nurses about IPT. To design the materials, we used prior findings from this province on patient barriers and facilitators to IPT [16]. Before the start of the study, we also conducted two formative focus group discussions with twelve total patients from the patient-centered intervention clinics to solicit feedback on initial drafts of the materials. The final patient-centered intervention incorporated this feedback prior to implementation.

Each clinic had two queues in separate spaces. One queue was for chronic care and the other queue was for maternal and child care. The median queue time in similar, neighboring clinics was 3.2 hours (interquartile [IQR]: 1.6 – 4.5 hours), with the vast majority of patients arriving in the morning (unpublished data). To educate all patients regardless of arrival time, we conducted two sessions for both queues, once in the morning and again in the afternoon, for a total of four sessions daily per clinic.

Health education sessions reviewed HIV symptoms, transmission, prevention, and treatment; TB symptoms, transmission, prevention, and treatment; differences between latent TB infection and active TB; purpose and benefits of IPT; and encouragement for patients to ask nurses about IPT. Assistants conducted the oral health education sessions in Setswana, Sesotho, English, and/or Xhosa depending on audience makeup and repeating in different languages as needed. Assistants distributed 2,839 calendars across 3,176 non-unique attendees, as some patients both attended and were counted in multiple sessions. Assistants offered all patients at least one calendar. In an average education session, research assistants spent 4.5 minutes teaching, 9.5 minutes engaging patients with questions and distributing calendars to those who responded, and 1.3 minutes distributing calendars to the remainder of patients at the end of each session.

Two South African research assistants conducted the health education sessions, aided by large visual depictions of key concepts. Both had Bachelor’s degrees in psychology and were specially trained on the epidemiology of TB and HIV, the health belief model [17], national guidelines for prescribing IPT [5], and audience engagement techniques.

Data collection

For the quantitative assessment of the study, the clinical mentor took attendance after interacting with nurses in workshops or in-room consultations. Research assistants recorded the number of patients who attended each in-queue education session.

For the qualitative assessment of the study, a female researcher from South Africa made in-person requests for interviews with all eligible nurses in nurse-centered intervention clinics. The qualitative researcher also made in-person requests for interviews with a convenience sample of patients who had just finished listening to the in-queue education session at patient-centered intervention clinics. We recruited patients until we reached data saturation [18]. The qualitative researcher conducted and audio recorded face-to-face, in-depth interviews with nurse and patient participants in local languages in which interviewees were fluent (English, Setswana, and Sesotho). The researcher administered a short demographics survey to participants and then used semi-structured interview guides to explore knowledge about and the acceptability of IPT. The qualitative researcher had no prior relationship with any nurse or patient and conducted interviews in a private clinic room. Immediately after each interview, the qualitative researcher summarized key findings into memos. Audio recordings were translated and transcribed into English, returned to the nurse participants to verify accuracy, and de-identified for data analysis. In the last month of the study, the lead author also conducted informal debriefing sessions with nurses at all four clinics.

Outcomes

There were four implementation outcomes: adoption, fidelity, acceptability, and sustainability. For the quantitative assessment, we measured adoption of the nurse-centered intervention and fidelity of the patient-centered intervention. Adoption is the uptake of the intervention, and fidelity is the extent to which the intervention was implemented as initially planned [19]. For the qualitative assessment, we gauged participant perceptions about the acceptability and sustainability of the interventions. Acceptability describes whether a stakeholder finds a practice to be “agreeable, palatable, or satisfactory,” and sustainability relates to whether a new behavior will be maintained during normal clinic operations [18].

Analysis

We calculated the adoption of the nurse-centered intervention as the proportion of nurses who attended the mentor-led workshops and consented to have in-room consultations. We measured the fidelity of the patient-centered intervention as the total number of education sessions performed as a proportion of the targeted four sessions per clinic for each of the 67 working days spent on site. Adoption and fidelity were considered low if below 40%, moderate if between 40–79%, and high if 80% or above, based on cut-offs from previous health education interventions from the implementation science literature in resource-constrained settings [20–23].

We assessed the acceptability and sustainability of the intervention using thematic analysis of the in-depth interviews [24]. Two research assistants coded the transcripts by hand with pre-specified codes (barriers to the targeted behaviors and what made each intervention acceptable or sustainable). The lead author then read the transcripts line-by-line, coded them with the same pre-specified codes in Atlas.ti (version 1.0.50), and open coded emergent concepts. The qualitative researcher, research assistants, and lead author discussed these themes to ensure that they were inclusive of all participant interviews. While re-reading coded extracts and interviews, the lead author further revised and refined codes into themes in a way that incorporated disconfirming evidence. Themes were iteratively refined in conversation with co-authors through grouping, drawing connections, and evaluating frequency and saliency. The lead author then checked the final themes by re-reading all coded extracts and interviews. The lead and second authors selected quotations primarily on the basis of representing key themes and using an iterative process of peer debriefing.

Results

From October 2016 through February 2017, we approached thirteen nurses and eleven patients to participate. We were unable to interview three eligible nurses due to scheduling conflicts; one patient was ineligible because of being HIV-negative. In total, we interviewed 10 nurses and 10 patients (Table II). Interviews were a median of 30 (IQR: 27–33) minutes with nurses and 15 (IQR: 15–19) minutes with patients. On average, nurses had been practicing at their current clinical level for 5.9 years (standard deviation [SD] = 4.1), had been working in their clinic for 4.6 (SD = 3.6) years, and were 37.2 (SD = 9.6) years old. Patients had been HIV-positive for an average of 5.5 years (SD = 4.2) and were 34.9 (SD = 11.7) years old.

Table 2:

Participant Demographics

Nurses
ID	Clinic	Years at Clinical Level	Role	Age
1	A	0.9	Community Service Nurse	25
2	A	11	Professional Nurse	42
6	A	0.5	Community Service Nurse	24
7	A	10	Professional Nurse	40
8	A	11	Professional Nurse	42
10	A	2	Professional Nurse	26
3	B	5	Professional Nurse	40
4	B	9	Nurse Manager	36
5	B	4	Professional Nurse	43
9	B	6	Professional Nurse	54
Patients
ID	Clinic	Years HIV-positive	Sex	Age
1	C	Not reported	Female	–
4	C	4.3	Female	25
5	C	9.6	Female	36
7	C	0.2	Female	39
10	C	4.1	Female	37
2	D	Not reported	Male	–
3	D	Not reported	Female	25
6	D	8.6	Male	57
8	D	1.1	Female	20
11	D	10.9	Male	40

Adoption and Fidelity

Adoption of the nurse intervention was high, with twelve of thirteen eligible nurses attending both training workshops (92.3%). Additionally, the clinical mentor provided a total of 38 in-room consultations across all 13 eligible nurses, each lasting an average of 80 minutes. Fidelity to the patient intervention was moderate, with research assistants providing 167 of the 268 (62.3%) planned education sessions.

Acceptability: Nurse Intervention

Overall, nurses described the intervention as acceptable and many expressed a desire for it to continue. Nurses particularly enjoyed the participatory aspects of role-playing patient scenarios, developing clinic-specific Action Plans, and learning about one another’s challenges in caring for HIV-positive patients. Nurses expressed that two aspects of the training were particularly critical for motivating change: first, the clarification of national guidelines and second, the external oversight from the clinical mentor.

Utility of Participatory Workshops to Clarify Ambiguous National IPT Guidelines

Nurses unanimously described the South African national IPT guidelines as ambiguous, confusing, and incomplete. As a result, a majority reported having no practical experience applying the guidelines in a clinical setting. For example, one nurse commented that although she had placed a tuberculin skin test (TST) at a regional training, she had never done the test on a patient in her clinic and was unsure whether she could read the result.

Nurses were also unsure whether they could initiate a patient on IPT without placing a TST and how the prescription duration — which can span from six months to three years — was impacted by patient age, pregnancy, antiretroviral status, and results from the TST. Nurses pointed out that although the national guidelines instruct them to prescribe IPT even if TST is out of stock, the guidelines also state that a TST must be placed one month after IPT initiation. The guidelines have no instruction of what to do if TST remains out of stock after IPT initiation. Nurses said that this feeling of uncertainty and ambiguity of guidelines sometimes prevented them from prescribing altogether:

Previously we were not sure… now there are [IPT durations of] three years… one year. We were all not sure for how long… We thought rather than making mistakes, let’s not give it out. But now [that we’ve completed the workshops] there is a change. (Nurse 007)

Speaking to the effects of the intervention, the same nurse explained this “change” at another point in the interview as “having light.” In this instance, “light” was a reference to gaining a more practical understanding of how to follow the IPT guidelines and how those guidelines fit into pre-existing clinical practice for HIV patients. In particular, this nurse said she and her colleagues were unsure how to determine the length of time that a patient should be on IPT.

Another nurse echoed this sentiment by explaining that the workshop had motivated her to prescribe IPT by providing a full understanding that included the how and why of IPT:

Interviewer: What is it that … is pushing [the other nurses] to initiate more IPT?

Nurse: We are doing something that we know. That you understand clearly. It can do wonders. I really like to do things that I know, especially nursing. Yoh! It’s not easy to just say, ‘I will… give [IPT…or just a medication] because they said we must give.’ That’s what happened previously. (Nurse 008)

This quotation reveals that this nurse also had difficulty complying with the national guidelines based on incomplete knowledge. She found the workshop useful because it addressed a gap in that knowledge while also going beyond to emphasize the importance of IPT for HIV-positive patients and hence helped change her behavior.

Further Benefit from External Oversight from the Onsite Clinical Mentor

In addition to describing the utility of clear information, nurses recounted motivation coming from a variety of sources. Some nurses, like the one quoted above, were intrinsically motivated to practice something they perceived themselves as fully grasping. Other nurses said the IPT Champion motivated them with encouragement and weekly announcements at group meetings. Most of all, nurses and the IPT Champions reported being most motivated by the presence of the clinical mentor. Nurses specifically appreciated that the clinical mentor did not chastise them or demand changes without reason. Instead, the clinical mentor gently emphasized looking at the protocols whenever there were deviations:

She will just do a check and say, ‘Oh sister, didn’t you forget this and this’ and then we go back to the protocol together. You’ll see that? There? This is the real nurse because the protocols are really actually what we are required to use. (Nurse 008)

Despite the importance of the clinical mentor, most nurses were initially uncomfortable with her observing their work. This anxiety subsided for all but one nurse, who stated that her work with patients was obstructed due to the clinical mentor’s presence:

The patients… are not used to her. When they see her in the room, some do not feel free… I have noticed that some of them when she is here, they are not open. I am mostly working with TB and HIV patients… But when I am with someone, they will be quiet, and that doesn’t give me the chance to acquire more information that I need… (Nurse 002)

When this nurse suspected that certain patients might not be willing to disclose relevant medical information in the presence of the clinical mentor, she asked to have in-room consultations on a different day. The nurse stated that she still consulted the clinical mentor for clarifications about IPT.

Acceptability: Patient Intervention

Patients approved of the education session content and said it was acceptable to provide HIV-related education to the entire queue, so long as it did not reveal their status to others. Patient interviews demonstrated that the intervention had heightened awareness around HIV and TB but did not effectively aid patients in retaining significant information about IPT. For instance, five out of ten patients — three IPT-naïve and two patients who had previously taken IPT — knew that IPT prevented TB disease. One of 10 patients could recall the full name of the therapy (i.e., isoniazid) or its acronyms (i.e., IPT or INH). After hearing the session multiple times, later patient interviewees demonstrated somewhat better recall, suggesting that repeated education sessions enhanced patient knowledge.

Beyond the sessions themselves, the calendar incentive for participating in the education session was particularly well received. Patients expressed appreciation for its depiction of a local monument and acting as a way for patients to remember their appointments. Every patient interviewed said they would hang the calendar in their home and would share what they had learned with their close family and friends, a sentiment captured by one participant who said:

There is no way that I cannot share it with others. It is also important to share for those who are not living with HIV as well. To share knowledge that TB is real, and you can live with it. You can be on treatment for certain periods and get well again. I will [also] share the light on TB symptoms… (Patient 002)

Despite an apparent willingness to hang the calendar in their home, no patient indicated full awareness of its content about HIV, TB, and IPT. In comparison to the calendars, no patient mentioned the posters unprompted, and when prompted, many said they had not seen the posters or were, as one patient said, “not paying any attention to them.”

Finally, though patients reported being more prepared to have a conversation with their provider about IPT, many patients stated they were still unsure of how to begin. In the patient-centered intervention clinics, discussion with nurses revealed that there had been few requests for IPT. A lack of requests indicated that despite patient acceptability of the intervention, barriers remained for patients to bridge the gap between intending and enacting the target behavior.

Sustainability

Both nurses and HIV-positive patients expressed that their intervention was overall acceptable and had provided them with sufficient knowledge about IPT. Despite this, almost every interviewee stated that the intervention alone could not achieve or sustain sufficient change. Two common reasons surfaced across interviews. First, nurses and patients stated that the individual is responsible for acting after being informed about IPT:

I think that you [outside researchers] can only do things up to a certain standard… the source of motivation comes from an individual self. So, there is not much you can do, with regard to trying to motivate somebody. (Nurse 001)

I don’t know exactly what is it that you can do [to help me talk with the nurses about IPT] … it’s for us to be equipped with the knowledge about IPT while we are at the clinic. From there, it is my duty to decide to go and tell the nurses that I need IPT that I heard about. (Patient 005)

Second, even when motivated to enact the target behaviors of asking about and prescribing IPT, patients and nurses described the opposite party as resistant to change. For instance, nurses said that patients misunderstood or were unwilling to cooperate with an IPT prescription, as exemplified by one nurse who said:

I have already received back the IPT from two of my clients… I gave them education and when the community health workers visited, [the patients] returned the IPT – saying that they don’t need it anymore because they don’t have TB… (Nurse 002)

Patients, in turn, said they were often intimidated during their conversations with nurses. Patients expected nurses to initiate and explain new medications as needed. Of the ten patients interviewed, one reported asking a nurse for IPT during an appointment. The patient described being incorrectly denied an IPT prescription because the nurse told them that they “did not have TB.”

The remainder of patients stated that they intended to ask about IPT but could not find a platform. Although patients said most nurses were approachable, nearly half reported that nurses were sometimes rude, and that this discouraged patients from asking about IPT. As one patient said:

Honestly, you get different types of nurses here. You get types that when [they] know that you are an HIV patient… some have like an attitude towards you. And then you will get some that tries to be nice to you. So, sometimes you don’t even know if you should ask the question or if you shouldn’t ask the question. It is like you are being inconvenient… If I see that you have an attitude, I won’t bother even asking. I’ll rather wait for someone else who is more willing to work with me and be friendly. Someone that I can see and can ask you questions because you really want to help me. (Patient 008)

In addition to the above barriers mentioned by both nurses and patients, nurses pointed to several clinic- and district-level barriers that would impede long-term change. First, in the absence of external oversight, nurses said that the IPT Champions would stop their work and nurses would lose their motivation to prescribe IPT:

I think you must keep on visiting us, keep on giving us information… Check whether are we still on the right track or are we doing the right thing. You shouldn’t just come once… and think that we are still doing it. (Nurse 002)

Some nurses said that the IPT Champion alone was not enough motivation and suggested that the effort might be better sustained with refresher trainings, support, and oversight from district-level management. Others suggested that text messages could mitigate a decline in motivation. For example, one nurse agreed the intervention could continue to be effective from afar:

They sent us… SMSes [so] we will continue to remember that there is IPT. We have to continue. If I miss some points during training, I’ll be able to know from those SMSes and scenarios that… ohhh! When the pregnant women come…. I have to do this and that. (Nurse 010)

This quotation demonstrates that the text message scenarios provided continued educational opportunities and were also a convenient touchstone for guidance on the appropriate care for different patient scenarios. In the absence of face-to-face clinical mentorship, the messages also reminded the nurses to continue practicing the national guidelines. Though text messages were generally found to be acceptable, nurses did not want to receive messages outside of work hours.

Other clinic-level barriers to sustained progress included IPT prescriptions outpacing incoming inventories of isoniazid and accompanying medications, resulting in district-wide stockouts. While Action Plans had arranged to increase IPT inventory at each clinic, the intervention had not considered the larger system. Nurses also frequently mentioned that disorganized patient charts and the lack of an IPT-specific log were significant barriers to patient continuity of care. Though there had previously been a district-wide IPT log, there had been recent national changes that removed them from the clinics. The intervention did not provide a way for nurses to easily track each patient’s progress toward their prescribed IPT duration, and without a log, most nurses stopped patients who had been initiated to avoid the possibility of prescribing IPT past the intended duration. Failure to maintain records also meant that data were not captured correctly, and improvements in IPT prescriptions were not accurately reflected at group meetings. Finally, nurses repeatedly stated that staff shortages and a large patient burden prevented them from screening every HIV-positive patient for IPT eligibility.

Discussion

This study assessed the adoption, fidelity, acceptance, and sustainability of two interventions for increasing the prescription of IPT to eligible HIV-positive individuals. The nurse-centered intervention had high adoption, and the patient-centered intervention was implemented with moderate fidelity. Nurses and patients both reported that the interventions were acceptable. Despite positive implementation outcomes, this study revealed that participants did not describe themselves as fully capacitated or motivated to sustain the behavior changes their interventions aimed to achieve. Furthermore, gains from the nurse-centered intervention were perceived as unsustainable and beyond each clinic’s capacity without district-level changes.

Both nurses and patients pointed to one another as reasons that IPT prescriptions were not being fully achieved, which is a common phenomenon in interventions requiring the cooperation of multiple parties and akin to self-motivated attributional biases [25, 26]. Both patients and clinicians in this and other studies in South Africa, though, generally perceived clinicians as the party controlling patient care [7, 27]. This squarely places the onus on clinicians to promote IPT and on the health system to provide unambiguous guidelines and effective training to clinicians.

Though patients in this study found the intervention acceptable, few enacted the targeted behavior of asking a nurse whether IPT was right for them. Patients described uncertainty in how to begin the conversation or whether to offer their opinion, indicating a lack of agency or perceived behavioral control over their care [28, 29]. To improve the implementation of IPT in the Northwest Province, patients suggested changing patient-nurse communication to empower the patient.

Patient-centered communication is one option for nurses to shift the power dynamic and enable patients to more actively engage in their care [30]. For example, nurses could communicate in a way that shares power and responsibility with their patients by eliciting patient ideas or expectations. This type of communication may encourage some patients to ask about IPT, but previous work in South Africa regarding IPT promotion indicates that many patients do not ask questions since they believe that providers act in their best interests [27]. This is bolstered by a separate study regarding patient preferences for communication styles in Mali, which has shown that although a majority of patients (40.2%) prefer sharing power with their provider, a large proportion (35.8%) still prefer physicians and other providers to lead while many have no preference (24.1%) [31]. Therefore patient-centered communication and education alone will not sustain IPT initiation. Proactive clinical action will be required.

Clinical education and training are essential to successful rollouts of any new medical guideline. The nurse-centered intervention in this study was moderately successful and motivated some nurses to perform the targeted behaviors. Nurses required intensive clinical training because the current IPT guidelines are neither easily nor quickly understood. Our data reinforced findings from a 2010 study in South Africa that uncertainty about the national guidelines and lack of experience prescribing IPT are major barriers to the uptake of IPT among clinicians [7]. Our study showed that an educational intervention may instill confidence in nurses to act if it (1) uses active learning techniques to introduce concepts, (2) shares the “light,” i.e. clarifies any ambiguities remaining after initial training sessions and emphasizes why IPT is important for patients living with HIV, and (3) reinforces learning using onsite clinical mentorship that both encourages action among hesitant nurses and supports the efforts of nurses already acting.

Nurses unanimously expressed concern, however, about the sustainability of the intervention and how they would maintain their new behavior. Similar to other nurse-education interventions to improve TB care elsewhere in South Africa [32–34] and sub-Saharan Africa [35], our study showed that onsite clinical mentorship may inspire short-term but not long-term change among nurses. Text messaging was one acceptable option for offering continued support. Such digital interventions, with a large reach but little effort, are especially promising in South Africa, which faces a severe shortage of healthcare workers including nurses [9, 10].

Beyond wanting continued clinical mentorship, nurses emphatically expressed a need for district-level initiatives to improve methods of stocking IPT sufficiently, monitoring individual IPT prescriptions month-to-month, maintaining reasonable nurse-to-patient ratios, and sustaining motivation to prescribe. These needs are similar to those reported from PALSA PLUS and NIMART, two successful training initiatives in South Africa that task shifted the diagnosis and treatment of TB and HIV to nurses in 2004 and 2011 respectively [36–38]. The success of these programs was heavily facilitated by district-level HIV and TB coordinators who mitigated system-level barriers. For example, during the PALSA PLUS rollout, coordinators initiated operational changes with clinic managers and found nurses to fill vacant positions [36].

This study had several limitations. First, the small number of clinics sampled may have prevented us from attaining sufficient levels of facility-specific variations for proper saturation of themes. Second, this work was conducted in clinics in townships, and the results may not be transferable to other regions of South Africa or the world. However, the rich description of findings makes the findings transferrable to other studies of nurse and patient perspectives on disease management in similar contexts. Finally, while interviewer effects (e.g. social desirability bias) may have influenced participant responses and analytic processes (e.g. confirmation bias) may have influenced the results, the following steps were taken to improve the credibility of these findings: writing memos after interviews, triangulation between nurse and patient participant results, peer debriefing, and nurse participant reviews of their transcripts.

Conclusion

Tuberculosis remains the single greatest cause of death among people living with HIV. Preventive treatment for TB is a crucial element for elimination, especially as case detection rates stagnate. This study described the first documented interventions to improve the provision of IPT to HIV-positive adults in South Africa. We evaluated both nurse and patient perspectives on the immediate and medium-term impacts of the interventions. Although nurses and patients found the interventions acceptable, nurses still required extensive training to overcome the ambiguity of the national IPT guidelines and patient education was not sufficient to motivate change. Complicated guidelines without adequate oversight or improved patient-provider communication will hinder preventive treatment implementation. Future guidelines should provide comprehensive instructions to cover all potential patient and medication stockout scenarios, be written in simple language, and be communicated with visual diagrams like decision trees.

Taken together, this study and others show that implementation barriers for IPT in South Africa will continue to persist without changes in both patient-provider communication and district-level action to increase support and monitoring for IPT. More research is needed to evaluate the effectiveness and sustainability of multi-level interventions that concurrently educate nurses and patients on IPT, provide sustained support to nurses through text messaging and district-level monitoring, and promote patient-centered communication.