Introduction
Integrating HIV care and treatment services into opioid treatment programs (OTP) can improve linkages to HIV care and antiretroviral therapy (ART) and optimize HIV treatment benefits for people who use drugs (PWUD) [
1]. However, there can be challenges specific to implementing care integration in setting with limited resources. Research indicates that adding or scaling up programs in these settings, can impact clinic efficiency and patient flow and without attention to structural barriers many patients might be lost to follow up [
2,
3]. In integrating care services, particularly in setting with limited resources and among vulnerable populations, there is a need to figure out how to deliver and sustain these efforts in ways that are effective, timely and of high quality. Simple translation of interventions, from one care context to another, may not address issues of cultural appropriateness, resource limitations, existing health care structures, or political will.
PWUD shoulder a disproportionate burden of HIV in Tanzania [
4]. HIV prevalence among PWUD living in Tanzania is estimated at 36% compared to 7% in the general population [
5,
6]. In an effort to address the high rates of HIV among PWUD in Tanzania, the first publicly-funded OTP on the mainland of sub-Saharan Africa opened in February 2011 at Muhimbili National Hospital (MNH) in Dar es Salaam. The MNH OTP clinic offers methadone maintenance treatment, as well as psychosocial and behavioral therapies, as part of its medication-assisted treatment of opioid use disorder [
7]. However, despite daily attendance at the MNH OTP clinic, less than half of treatment-eligible patients initiated ART within 3 months of being deemed eligible for treatment [
8], due partly to delays in CD4 testing, suboptimal systems to monitor patients and link them to care, and siloed health care structures [
9].
To address delays in ART initiation and improve HIV-related clinical outcomes among PWUD attending the MNH OTP clinic, we implemented the Integrated Methadone and Anti-retroviral Therapy (IMAT) intervention at the OTP clinic starting in 2015. At the launch of IMAT implementation, the IMAT intervention included four key components: (1) in-house point-of-care CD4 testing; (2) in-house HIV clinical management by methadone clinic providers trained in comprehensive HIV management, with referrals to the HIV clinic for developing needs; (3) ART delivery through the OTP clinic; and (4) an electronic information system to help providers monitor OTP patients along the continuum of HIV care. At the beginning of 2016, the Tanzanian Ministry of Health and Social Welfare adopted a ‘test and treat’ model for HIV among PWUD at the OTP clinic. Following this change more patients were eligible to initiate onto ART due to the elimination of reliance on a specific CD4 threshold to determine ART eligibility. Since this change occurred after IMAT it has effectively enabled providers at the OTP clinic to operationalize the first ‘test and treat’ model for HIV among PWUD in sub-Saharan Africa. In this paper, qualitative data were collected to examine provider and patient perspectives on the implementation of integrated methadone and HIV services at the Muhimbili OTP clinic.
Methods
Semi-structured in-depth interviews were conducted with 35 HIV-positive OTP patients and 8 OTP providers and at the MNH OTP clinic 6-months after IMAT implementation. We interviewed providers at the OTP clinic who were involved with IMAT at the time of data collection, which included 3 nurses, 2 doctors, 1 pharmacist, 1 social worker and 1 administrative person. Providers were asked about their reaction to and opinions of the IMAT intervention including its implementation, their role in patient education, intervention procedures, and ART dispensing.
The 35 patients interviewed were purposively sampled based on sex and ART treatment status, and for those on ART, when they were linked to ART (Table
1). Interviews with patients focused on their experiences with the IMAT intervention and adapting to the new protocol. Patients were eligible for the study if they were HIV seropositive and enrolled in care at the OTP clinic at the time of data collection. Providers at the clinic determined if patients met eligibility requirements for in-depth interviews. Those who were eligible for study participation were asked if they were interested in participating during their private appointments, as to limit the risk of HIV status disclosure. Interviews were conducted in a private location and were audio-recorded, transcribed in Swahili, and then translated into English. This study received ethical approval from the Tanzania National Institute for Medical Research, the Muhimbili University of Health and Allied Sciences Institutional Review Board, and Ethical and Independent (E&I) Review Services in the United States.
Table 1
Characteristics of HIV-positive OST patients interviewed for the study (n = 35)
HIV positive OTP clinic patients not on ART, post IMAT | 1 | 2 | 3 |
HIV positive OTP clinic patients not linked to ART before IMAT, now on ART | 23 | 9 | 32 |
Total | 24 | 11 | 35 |
Data analysis
Interview transcripts were entered into Dedoose (Version 7.0.23) for storage, organization, coding and analysis. The CFIR was used to guide data coding and analysis. Memos were used at each stage of data analysis to saturate analytic categories and facilitate analysis.
The CFIR framework includes five domains: intervention characteristics, outer setting, inner setting, characteristics of individuals involved and process [
10]. In this study, we applied these domains to understand patient and provider perspectives on the context in which the intervention was delivered, rather than individual outcomes. Using available data, we focused on three out of five CFIR domains: intervention characteristics, outer setting, and inner setting. We did not apply the domain of process in this study because we were examining implementation at one site, and did not have data on quality and extent of planning, engagement of key stakeholders and did not conduct a multi-site comparison of health care delivery systems. Additionally, we did not apply the domain of characteristics of individuals involved because this domain is used to understand the behavior of the implementer or participant and its effect on implementation [
11]. Of the CFIR constructs assessed, available data resulted in operationalization of eight implementation constructs: complexity, adaptability, relative advantage, evidence, patient needs and resources, available resources, and compatibility. These constructs and their operationalization, as defined by Damschroder [
10] are presented in Table
2.
Table 2
Implementation constructs
Intervention characteristic | Complexity | The complexity (in terms of time, steps, or difficulty) of the intervention itself |
| Adaptability | The ability, or lack thereof, to adapt the intervention to the specific clinic context |
| Relative advantage | Improvements or worsening by the intervention, compared to currently existing programs, or having nothing in place of the intervention |
| Evidence | Stakeholders’ perceptions of the quality and validity of evidence supporting the belief that the innovation will have desired outcomes |
Outer setting | Patient needs and resources | The extent to which the intervention addressed or accounted for patient needs and limited resources |
Inner setting | Available resources | The presence or absence of resources specific to the intervention |
| Compatibility | The level of compatibility the intervention had with work processes and organizational values |
Each interview was coded by the first author (AC), and these codes were used to develop case memos for each construct. Codes related to the construct of patient needs and resources included statements that discussed awareness, or lack of awareness about the needs and resources of those served by the intervention. Codes related to the construct of available resources included statements related to the presence or absence of resources specific to the intervention. Codes for compatibility included statements that discussed the level of compatibility the intervention had with work processes and organizational values. Statements were coded for the construct of complexity if they discussed the complexity of the intervention itself. Codes related to the construct of adaptability included statements regarding the ability, or lack thereof, to adapt the intervention to the specific clinic context. Codes for relative advantage included statements demonstrating that the intervention was better or worse than existing programs, or having nothing in place of the intervention. Lastly, codes for evidence included statements related to providers’ perceptions of the quality and validity of evidence supporting the belief that the innovation will have desired outcomes. Memos contained a summary of all relevant codes for the construct, a rationale for the code and the direct quotations related to the construct. From these memos we were able to assess the construct’s influence on IMAT implementation.
Discussion
In this paper we have used the CFIR to understand contextual barriers and facilitators related to implementation of the IMAT intervention. In-depth interviews with patients and providers 6-months post-implementation discussed how intervention characteristics, inner setting and outer setting factors influenced intervention implementation. Understanding the contextual factors that influence implementation is critical to the dissemination, scale-up and adaptation of the IMAT intervention to other settings.
Assessing context is important when evaluating interventions because it takes into account the fact that implementation outcomes neither exist nor arise in a vacuum. Context refers to the set of circumstances or unique factors that surround a particular implementation effort, and which must be accounted for in data interpretation [
12]. Doing so allows us to assess whether intervention success or failure is related to problems with the protocol, or to the intervention occurring in a context without adequate resources, support, and the systems necessary for success. This is in alignment with implementation science literature that posits a clinic’s capacity to conduct an intervention does not necessarily reflect whether the intervention itself is effective or ineffective, but rather whether the intervention is implemented effectively and as intended in a particular context [
13]. As applied to this study, we see differences between patients and providers, related to how context impacted intervention implementation. While the IMAT intervention was able to address patient needs in terms of ART initiation, more contextual factors such as food insecurity and transportation exist outside of the clinic’s purview, but still impact patient treatment and care. While these contextual factors may not impact the number of clients initiating ART it will certainly impact their medication adherence and therefore rates of viral load suppression which is a critical determinant of HIV related morbidity and mortality. Similar interventions might be able to facilitate implementation and create a more supportive environment through the expansion of services, allowing care to be located closer to where patients live (thus minimizing travel burden) and linking patients to relevant services, such as food assistance programs.
Facilitators of IMAT implementation included the CFIR domains of adaptability, relative advantage and evidence. These are all attributes of the intervention that were directly addressed by stakeholders (OTP patients, OTP providers, Local NGOs delivering HIV care services, HIV experts) at the time of intervention design [
14]. This group identified ART initiation as a problem that could be addressed by integration of services. Furthermore, they highlighted the importance of designing an intervention that allowed for adaptability to meet the needs of providers (e.g. flexibility) and patients (e.g. confidentiality) as critical to the success of a complex intervention. The post-IMAT interviews confirmed that incorporating these stakeholders’ views provided for key facilitators for implementation. While not part of the intent of IMAT, interviews with patients and providers also suggest that integrating HIV care at the OTP clinic helped to improve retention as well. In looking to implement a similar integration program in other settings, it is important to incorporate stakeholder priorities in the intervention design to facilitate implementation.
The most significant barrier to implementation identified in interviews centered on contextual factors related to available resources. For some providers, lack of available resources (e.g. space) significantly hindered implementation and intervention delivery. While the shift to a ‘test and treat model’ was not explicitly discussed in interviews, it did result in increasing the number of OTP patients eligible for ART initiation. While providers discussed challenges related to available resources this was not a barrier endorsed in patient interviews. In interviews, IMAT patients felt they were able to easily receive services, which may indicate that IMAT patients are not acutely awareness of limitations in clinic resources or that providers were able to manage them without impacting their clinical care. In limited resource settings, prior service integration efforts in other health sectors have also shown the importance of policy-level efforts to improve the efficiency in resource utilization (providers and financial) of integration efforts [
15]. These issues of available resources and compatibility will need to be addressed at the policy and service-delivery level, especially in context of a ‘test and treat’ model to ensure the sustainability and scalability of IMAT as well as other integration of service efforts.
In provider interviews we observed a discrepancy in comments about compatibility compared to comments made about available resources. While providers felt it was easy for them to incorporate intervention procedures into existing workflows and systems at the clinic, they also indicated that the level of resources (e.g. workforce, space and lab equipment) dedicated to implementation had an overall negative influence on implementation. The strain on resources may also been seen as a sign of successful implementation. Following the shift to a ‘test and treat model’, there was a sudden increase in the number of patients to manage. Had IMAT not been compatible with the OTP clinic, we might observe patients dropping out or leaving care thus not straining clinic resources in this way. This suggests that there may be tension between the compatibility of an intervention for a particular setting, and the necessary tools and resources needed to implement the intervention effectively. Having an adaptable intervention will allow for modifications to the intervention that will allow for increased compatibility for the available resources.
Though this study did not examine individual outcomes following IMAT implementation, we should note that OTP providers conducted over 100 CD4 tests within the first 3 months of IMAT implementation. Nearly 40 clients were seen by HIV care and treatment-trained clinicians for ART initiation at the MAT clinic. Following this the OTP clinic adopted a ‘test and treat’ model for HIV among PWUD at the OPT clinic, thus eliminating reliance on CD4 testing to determine ART eligibility and increasing workload demands [
16].
There are limitations of this study. We were not able to the domain of process in this study due to lack of available data. Multi-site comparisons of intervention delivery may be a key component in understanding and assessing how implementation is carried out, as well as potential intervention barriers that occurred as a result of the implementation process. It is possible that the exclusion of this construct limits perspectives of implementation barriers and facilitators. We only interviewed patients currently enrolled and providers working at the MNH methadone clinic. Patients who have defaulted or who were not currently enrolled in care for other reasons may have very different perspectives around these issues. Thus, patient interview data may be biased in only collecting the experiences of those who have been successful in treatment. In addition, data for this study were collected in one setting, raising questions around generalizability.
Authors’ contributions
AC analyzed interview data using thematic context analysis. Manuscript preparation was divided between AC, HS and SH. Feedback and editing was provided by DM JM and BL, who contributed greatly to manuscript development. All authors read and approved the final manuscript.