Article Text
PDF
Abstract
Objectives To assess the feasibility considerations for a decentralised, one-stop-shop model of care implemented in Yangon, Myanmar.
Setting Two primary care level clinics in urban Yangon, Myanmar.
Design This is a feasibility study of a highly effective care model. Using Intervention Complexity Framework by Gericke et al, we collated and analysed programmatic data and evaluation data to outline key project implementation requirements and experiences.
Participants Programmatic data were collected from clinical records, GeneXpert device test and maintenance reports, national guidelines, product and device instructions and site monitoring visit reports. Healthcare providers involved in delivering care model contributed interview data.
Results The main feasibility considerations are appropriate storage for test kits and treatments (in response to temperature and humidity requirements), installation of a continuous stable electricity supply for the GeneXpert device, air-conditioning for the laboratory room hosting GeneXpert, access to a laboratory for pretreatment assessments and clear referral pathways for specialist consultation when required. Lessons from our project implementation experiences included the extensive time requirements for patient education, the importance of regular error monitoring and stock storage reviews and that flexible appointment scheduling and robust reminder system likely contributed to high retention in care.
Conclusions Detailed documentation and dissemination of feasibility requirements and implementation considerations is vital to assist others to successfully implement a similar model of care elsewhere. We provide 10 recommendations for successful implementation.
Trial registration number The trial was registered at ClinicalTrials.gov NCT03939013 on May 6, 2019. This manuscript presents post-results data on feasibility.
- public health
- gastroenterology
- health services administration & management
- international health services
- public health
Data availability statement
No data are available. No unpublished data available for sharing.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
- public health
- gastroenterology
- health services administration & management
- international health services
- public health
Strengths and limitations of this study
This feasibility study is one of the first studies from a low-income or middle-income country to document the operational considerations of implementing a decentralised testing and treatment care model for hepatitis C.
Key strength of this study is that while it was a research study, the model was partially embedded in the healthcare system allowing for identification of scalability challenges.
Another strength is the use of various data sources for the evaluation of the feasibility of this care model.
Key limitation of this study is that external funding covered modifications required to overcome key implementation challenges, namely laboratory set-up.
There is the risk of response bias in the healthcare staff interviews as interviews were conducted by study coordinator.
Introduction
An estimated 58 million people live with chronic hepatitis C virus (HCV) infection worldwide, 75% in low-income and middle-income countries (LMICs).1 2 In 2016, the World Health Organization (WHO) targeted HCV elimination by 20303; however, few countries are on track to achieve it,4 and 80% of people living with HCV lack access to treatment.2 Highly efficacious pan-genotypic direct-acting antiviral (DAA) regimens (rendering genotyping unnecessary) and simple non-invasive liver fibrosis assessment using aspartate/platelet ratio index (APRI) simplify HCV clinical pathways and allow for adoption of a ‘treat all’ approach.5 6 The ‘treat all’ approach, now recommended by WHO, refers to offering DAA treatment to all persons diagnosed with HCV5; previously, only assessment for treatment was recommended given the complexities and lower efficacy of the previous interferon-based treatments.7 Coupled with diagnostic point-of-care (PoC) testing, these advances enable decentralised care. Decentralisation of care from central tertiary hospitals into district-level hospitals, primary care clinics or harm reduction sites is important for increasing access to care, partly by increasing geographical coverage of care sites. The unaffordability of diagnostics and DAAs and insufficient financing of national HCV programmes continue to prevent elimination8 and necessitate finding more cost-effective ways to provide care in resource-constrained settings.
Further scale-up of HCV testing and treatment in LMICs is needed, in turn requiring affordable, simplified HCV clinical pathways. In 2018, WHO endorsed eight good practice principles for simplified HCV service delivery.5 Creating and evaluating such models of care is vital to achieve elimination.6 Evidence supporting the effectiveness of decentralised HCV care models and task-shifting is growing, although with gaps in data from LMICs.9 A systematic review of 142 studies showed that full decentralisation and integration of HCV testing and treatment into harm-reduction sites or primary care improved access to testing and treatment.9 Moreover, task-shifting of treatment to non-specialists produced cure rates similar to those from specialist care, across multiple populations and settings.9 Decentralised HIV models have already shown how community-based interventions improve retention in HIV care and clinical outcomes in LMICs.10–12 Now, we need operational research evidence to guide implementation of decentralised HCV models of care, simplified clinical pathways and task-shifting. Evidence of such programme implementation and effectiveness will inform the planned 2021 update of WHO’s HCV guideline recommendations for simplified service delivery and use of PoC HCV viral load (VL) testing.
To assess feasibility and implementation considerations to inform future models and scale-up, we evaluated a simplified, ‘one-stop-shop’ HCV model of care in Yangon, Myanmar.13 14 Myanmar (population 53 million) is a South-East Asian LMIC15 with a mixed epidemic of HCV, with 2.7% anti-HCV positivity in the general population and 56% among people who inject drugs (PWIDs),16 and an estimated million people living with HCV. Access to HCV care is limited by the number of hepatologists (estimated at 25 nationwide) and unaffordability of diagnostics and DAAs. The national treatment programme has treated approximately 11 000 people living with HCV,17 but considerable scale-up and expansion is required to reach all affected.18 Fortunately, general practitioners (GPs) can prescribe DAAs and national guidelines also support simplified pretreatment assessments (no genotyping or FibroScan).19 Creating and evaluating a simplified clinical pathway and model of care for Myanmar is an important step towards expanding decentralised primary care-based HCV testing and treatment.
In this paper, we outline the implementation requirements of our simplified HCV model of care and our experiences in implementing it in Myanmar.
Methods
Study design
We aimed to assess the implementation requirements and feasibility considerations for our model of care and potential scale-up. We assessed the feasibility of the HCV testing workflow (delivering PoC serological rapid diagnostic tests (RDTs) and onsite PoC HCV RNA GeneXpert tests for VL confirmatory diagnosis) and treatment workflow (GP-led prescription of DAAs).
To guide the assessment, we adopted indicators (table 1A and B) from Intervention Complexity Framework (hereafter, Framework) by Gericke et al 20 and used mixed methods to collect data on study implementation and evaluation (including interviews with healthcare providers) and routine programme data. The Framework categorises health interventions by technical complexity (the quality and quantity of non-financial resources required to implement and sustain an intervention) to assist prioritising and scaling up interventions in resource-constrained settings.20 We used the Framework to analyse the implementation requirements and feasibility considerations of the care model, using all data sources.
(A) Evaluation of intervention complexity (HCV testing workflow); (B) evaluation of intervention complexity (HCV treatment workflow)
Description of clinical model of care
We trialled a decentralised ‘one-stop-shop’ model of care for HCV at two community clinics in Yangon between January 2019 and August 2020. Its defining features are a simplified clinical pathway (single rapid test for HCV antibodies, no HCV genotyping, use of blood-based biomarkers only for fibrosis assessment), community-based sites (no tertiary-level hospitals) and onsite phlebotomy and PoC testing for HCV diagnosis, minimising patient visits.
Clinical study procedures13 and outcomes are detailed elsewhere; for clarity, this study was a feasibility trial with no comparator arm.13 14 In brief, a laboratory technician took venous blood at patient enrolment; if rapid anti-HCV test (SD Bioline) was positive, reflex PoC HCV RNA testing (GeneXpert) was conducted on-site (no external laboratory confirmatory testing was performed for HCV diagnostic testing). After collecting baseline information and performing a physical examination and liver staging, the study GP initiated pan-genotypic DAA therapy (generic sofosbuvir, 400 mg; daclatasvir, 60 mg) according to national guidelines at the second visit for those with active HCV infection and eligible for treatment within the study. Treatment course length was determined using APRI score, as per national guidelines. Participants with physical signs of decompensation were referred to hepatologists, and those with significant comorbidities to other specialists, prior to DAA initiation. Drug dispensing and assessment of drug adherence and side effects occurred every 4 weeks. Figure 1 summarises the clinical steps and data collection relevant to this feasibility study.
CT2 feasibility—workflow and data collection. APRI, aspartate/platelet ratio index; DAA, direct-acting antiviral; GP, general practitioner; HBV, hepatitis B virus; HCV, hepatitis C virus; LFT, liver function test; RDT, rapid diagnostic test; SVR, sustained virological response; VL, viral load.
Outcome results are published elsewhere.14 In brief, we found very high retention in care (100% (606/606) of antibody-positive patients underwent RNA testing) and uptake of treatment (99.8% (488/489) among those eligible), and high treatment completion rates (98%, 477/488). We also found high cure rates in the whole participant group (92%; 421/456), including among PWID site: 91% (146/161) and people with compensated cirrhosis (83%, 19/23), all treated by trained GPs in the community. Median time from RNA testing to treatment initiation was 3 days (IQR 2–5).14
Description of implementation context
Setting
Yangon is Myanmar’s largest city.21 Relevant to stock storage, minimum daily temperatures are 17°C–25°C and maximums 29°C–37°C.22 Humidity ranges from a dry season average of 62% to 86% in monsoon season.22 Frequent scheduled and unscheduled power outages affect health service delivery.23
Sites and human resources
The two study sites were standalone clinics in residential streets. Each clinic had small onsite generators for power back-up. The Burnet Institute (BI) Thingangyun Key Population Service Centre served mainly PWIDs. It is a two-story apartment, equipped to meet clinic criteria, with a waiting area, consultation rooms, blood collection room and laboratory. The Myanmar Liver Foundation’s (MLF) Than Sitt Charity Clinic is a well-established clinic serving patients with liver disease due to hepatitis B or C infection. The laboratory was renovated 11 months into the study to improve the set-up for the GeneXpert device; see online supplemental appendix 1 for photos (laboratory set-up and schematic flow).
Supplemental material
Patients per month varied by site and recruitment phase—new participants and follow-ups, or follow-ups only. Median patients/month were 44 at BI clinic (IQR 3.5, 93.25, range: 0–209) and 75 at MLF clinic (IQR 14, 110.25, range: 1–429).
Each clinic was staffed by a GP, nurse, and laboratory technician. The BI site also had a peer worker for community outreach and sterile needle and syringe distribution.
Training was provided over 5 days to staff prior to project commencement, and then on-site training for the GeneXpert, quality assurance training and for the electronic medical records system—OpenMRS.
Further details on human resourcing and training are provided in online supplemental appendix 2.
Supplemental material
Data collection
We used semi-structured interviews, GeneXpert test and maintenance report data, medical records and document reviews for this feasibility analysis, summarised in figure 1.
Healthcare provider interviews
At 14 months into study implementation, when most participants had completed treatment and were returning for sustained virological response (SVR) testing, we conducted semi-structured interviews with all six healthcare providers to understand their perceptions of the feasibility of community-based hepatitis C testing and treatment (figure 1, evaluation stage). The six healthcare staff—a GP, laboratory technician and nurse from each clinic—were invited to participate and assured of confidentiality, and gave written informed consent.
Our interviewer (WLY, who has qualitative research training and experience) asked about the healthcare providers’ demographic characteristics, previous experience with laboratory procedures, GeneXpert testing and HCV treatment, advantages and disadvantages of community-based treatment versus hospital-based care and referral, logistics and storage procedures (the interview guide is shown in online supplemental appendix 3). Interviews were conducted face-to-face in Burmese in private at the clinics, audio-recorded and took about 1 hour. They were transcribed verbatim and translated into English by external consultants.
Supplemental material
Operational data
To inform scale-up and task-shifting opportunities, we collated the following operational data: GeneXpert test and maintenance reports, time for providers to complete tasks and review of relevant guidelines and product documentation.
We assessed GeneXpert device and operator performance using the following indicators: test outcomes (positive/negative, error, invalid), reason for error and maintenance schedule and outcomes (figure 1, visit 1, visit 6, evaluation stage). For a subset of participants, technicians collected timepoint data for laboratory procedures, including date and time of blood sample collection for HCV RDT and Xpert HCV VL testing, test start time, test read time and time to deliver test result; these were combined with treatment initiation date data to determine time to progress through the cascade. Time to complete specific clinical workflow tasks was recorded on 23 selected days (June 2019 to October 2019) across the study implementation period (to capture the recruitment, follow-up and SVR stages) as part of ingredients-based costing (figure 1), noting that no significant changes to clinical workflow occurred within the study period.
GeneXpert device requirements and instructions, DAA and RDT product inserts that provide information on its storage requirements, Myanmar national guidelines, and Foundation for Innovative New Diagnostics’ site monitoring visit reports were reviewed and contributed to the Framework assessment.
Patient involvement
Patients were not involved in study design, implementation or dissemination; however, 29 patients were interviewed to assess the acceptability of the model (findings are reported elsewhere24).
Data analysis
Quantitative data
GeneXpert HCV VL testing data (RNA testing for confirmation of viraemic infection for evaluation of treatment outcome and SVR 12±24 weeks post-treatment completion) from 30 January 2019 to 31 August 2020 were analysed for error frequency and error type by site. GeneXpert testing data were collated from OpenMRS and Xpert test log reports in Excel and imported into Stata V.15 (StataCorp) for cleaning and analysis. We calculated frequencies of tests and errors, error categories and codes and total testing days, by site.
GeneXpert maintenance reports were reviewed and key information summarised, including number of module replacements.
Qualitative data
Qualitative data, including interviews and documents, were managed and analysed using NVivo V.12 (QSR International, 2018). WLY and BLD independently conducted thematic analysis, using combined deductive coding based on the Framework and inductive coding for implementation considerations, matched to framework domains where applicable. WLY read and coded the healthcare provider interview transcripts in Burmese, and BLD read and coded the translated transcripts (English) to identify patterns and meanings independently. WLY and BD reviewed the initial codes together, discussed minor discrepancies and reached consensus on an initial coding frame; no major interpretative differences were raised. Codes with similar meaning were grouped into subthemes within framework domains.20
Data availability
No unpublished data available for sharing.
Results
Evaluation of intervention complexity: requirements for implementation of the HCV care model
The requirements for implementing this decentralised one-stop-shop model of care for HCV are presented in table 1A and B. We assessed two specific model components: testing workflow (table 1A) and treatment workflow (table 1B). Detailed version of table 1 (including table 1A and B with further detailed information) is included as online supplemental appendix 4.
Supplemental material
Project implementation experiences
Below we present experiences of project implementation, highlighting specific aspects that may be applicable when implementing similar models of care elsewhere. These data are mostly from provider interviews, and include data examining time to complete tasks and time to progress through care cascade.
Appointment scheduling and patient follow-up procedures
The healthcare providers described adapting appointment scheduling to the needs and convenience of patients (eg, allowing daily methadone dosing before appointments), and developed a reminder call system for upcoming appointments. Patients were generally asked to attend twice for dispensing after initial prescription (weeks 4 and 8). Staff mentioned that some patients would prefer to attend the clinic for DAA dispensing less often; future programmes could consider removing weeks 4 and 8 dispensing and provide a full course of DAAs at visit 2 to reduce follow-up appointments, in line with national guidelines.
Staff reminded all patients of appointments, and spent considerable time following up non-attenders. They reported some difficulties in reaching patients by phone, even when calling secondary contacts. If after multiple attempts the patient could not be reached, then the staff stopped calling but allowed them to continue with study appointments if they presented for care.
Recruitment targets for clinical workflow and workload
The number of new patients recruited daily was the key driver of healthcare provider workload. Staff set daily targets, adjusted weekly to reflect progress towards the overall target and the number of follow-up patients scheduled.
Staff described pressure to meet recruitment targets within the project’s timeline. Recruitment was paused in April until mid-June due to delayed DAA treatment procurement. When recruitment resumed, it was accelerated to reach the original target; this increased the staff’s workload and stress. Moreover, the four-module GeneXpert device meant that only eight samples could be run per day (the clinics operated 09:00–17:00 hours or 08:30–16:00 hours, unlike external laboratories, which have extended hours).
If we want to enrol more patients, like twelve to sixteen patients per day, we have to think about whether it is possible to do that here or collaborate with an external laboratory or run two GeneXpert machines simultaneously. (HCW05)
Recruitment capacity was reduced further when the modules required maintenance or replacement. Typically, five to eight new patients were recruited per day. Staff preferred steady enrolment. Generally, staff found the work manageable unless extra administrative and logistical tasks arose.
Managing workflow
The GP, nurse and laboratory technician combined to manage patient flow, with multiple patients seen at once at different stages of the workflow. Each healthcare provider had assigned tasks, but assisted each other as required, in particular with patient education and phlebotomy. Specific tasks were completed within each appointment, with some flexibility in order and timing to fit patient needs and clinic capacity.
HCV testing workflow performance
The median time from anti-HCV RDT to VL result delivery to participant was 159 min (IQR 140–430; n=154). Median GP time required was 10 min (IQR 8–12), nurse 30 min (IQR 16–37) and laboratory technician for initial study appointment, covering screening and diagnostic testing, 8 min (IQR 6–12). The nurse generally conducted pre-test counselling, and the GP conducted post-test counselling and medical history taking to prepare for treatment initiation.
HCV treatment workflow performance
For the second study appointment, when treatment was prescribed, median GP time was 12 min (IQR 8–16) and 4 min (IQR 1–15) for the nurse, covering review of pretreatment assessments, treatment prescription and dispensing and DAA education. Time with the nurse varied by site, dependent on workflow: at the BI site, median nurse time was 18 min (IQR 14–20) and at MLF site, 1 min (IQR 1–3).
The median time from RDT to DAA initiation was 3 days (IQR 2–5). There was a statistically significant difference between median time for the 95% of participants who did not require specialist review (3 days, IQR 2–5) and the 5% who did (20 days, IQR 15–36, p<0.001).
The most time-consuming aspect of consultations with healthcare staff was the patient education session. An education session was conducted after patients consented to participate. Consent and patient education procedures took up to 45 min. For patients with time constraints or concentration difficulties, staff covered the main points to save time. Staff refrained from disciplining patients because they focused on providing acceptable and accessible care to key population groups, as this example shows:
They keep talking on the phone and receiving calls during counselling. So, I had to let them talk because I cannot stop them … it was a bit difficult, but not that difficult. (HCW01)
In addition, if many patients were waiting, counselling sessions were sometimes shortened to allow subsequent clinical procedures to be undertaken and to ensure pretreatment blood testing was completed within the 1 day. A healthcare provider from MLF clinic suggested a private room, enabling one-on-one education without other staff or patients present would be beneficial.
Patient education
Healthcare providers reported that, despite repeated education and information provision, some patients misunderstood the modes of HCV transmission and the difference between HCV antibody tests and HCV RNA tests. Staff reflected that patients’ level of knowledge was related to their education, their interest in HCV and the general population’s poor access to mainstream health information. Staff suggested displaying printed resources at the clinic, using visual aids such as colourful flipcharts with pictures, and distributing brochures with key messages to facilitate health education.
Online supplemental appendix 5, table 2 details the number and types of errors experienced during the study implementation period. The BI site experienced an error rate of 3% and the MLF site 6%. The most common error at the BI site was ‘user/procedural’ (6/13, 46%), and at the MLF site, ‘other errors’ (17/42, 41%), followed by user/procedural (11/42, 26%) and cartridge errors (8/42, 19%). User/procedural errors generally stem from incorrect sample preparation.
Supplemental material
Stock storage
MLF staff struggled to keep the test kits and drugs below <30°C, especially when power outages occurred when the clinic was closed.
I have to push the staff a bit. For example, if the items and medicines are stored upstairs, I need to remind the staff to turn on the air-con. They are a bit forgetful like this. If I don’t push them, they will forget to do it. I always remind them. (HCW05)
Delivering GP-led care and referral pathways
Most patients who required hepatologist review were referred to MLF with few difficulties, but referral to specialists at public hospitals for comorbidities was challenging. Staff said patients often complained about long wait times at public hospitals. In addition, patients from the BI site told staff that early appointments meant they missed their morning methadone dose. The GP at one site described trying to avoid referring patients due to such difficulties. GPs suggested referrals to private clinics instead, but costs would be significantly higher than at public hospitals.
Referrals of hepatic complications are all right. It is a bit difficult to send [patients] to other specialists. The hard part is the queue time at the hospital … They cannot wait for that long… (HCW05)
Implementing an electronic medical records system
Patients’ clinical information was collected using OpenMRS. GPs preferred the electronic system (OpenMRS) over a paper-based system because as the electronic records were easily backed up and accessible. Staff reported that initial training on OpenMRS should include hands-on practice. Database field errors were common initially, but most were resolved quickly. Staff reported difficulty in resolving connectivity issues and errors without IT or technician support.
Discussion
This feasibility study showed a simplified HCV clinical pathway can be implemented in primary care facilities in Myanmar. Previous literature supports the safety and effectiveness of decentralised and task-shifting models of care, but few address operational considerations relevant to implementation and scale-up.9 Here, we provide new evidence about implementation and scale-up considerations for simplified, decentralised HCV models of care in LMICs.
We identified specific infrastructure requirements for successful HCV care, including continuous, stable electricity supply; adequate supply and stock management of Xpert cartridges; ongoing maintenance of the Xpert device; access to a laboratory for pretreatment assessment serology and established specimen referral logistics. The model was delivered by a trained GP; a laboratory technician or health worker trained in RDTs, phlebotomy and GeneXpert; access to a hepatologist for advice on suspected decompensated cirrhosis or complex cases and access to other specialists for comorbidities assessment. These were deemed the minimal requirements to deliver the HCV care model with fidelity.
We identified key operational challenges relating to laboratory set-up and procedures and successful completion of referral pathways, and propose solutions in the recommendations as mentioned in figures 2 and 3.
Key recommendations (recommendations 1–6). GP, general practitioner; HCV, hepatitis C virus; LMICs, low-income and middle-income countries; PoC, point-of-care; VL, viral load.
Key recommendations (recommendations 7–10). DAA, direct-acting antiviral; GP, general practitioner; HCV, hepatitis C virus.
HCV testing workflow
Phlebotomy
Standard phlebotomy was required for all patients for HCV PoC testing, liver function tests, full blood examination, renal function and cirrhosis assessment via APRI.
GeneXpert fingerstick assays for HCV diagnosis are simple, portable and require little equipment, increasing access and reducing test turnaround times. However, ensuring staff are well trained to collect blood from patients with difficult venous access remains important when engaging PWIDs as phlebotomy for assessments prior to DAA therapy is still required.
2. Laboratory set-up and GeneXpert errors
Appropriate laboratory set-up was important for Xpert HCV VL testing. The total error rate, 5%, was slightly higher than the 1%25 and 2%26 reported in other studies in community settings. However, a field validation study in Tanzania27 reported an error rate of 20% for HCV VL assays (plasma samples), possibly due to samples’ exposure to high temperatures, humidity and movement for 10 min during transport to the onsite laboratory.27 It is unclear exactly why our error rate was higher than other studies which report this, but it is likely due to a combination of factors including: the conditions of the room (ie, humidity, frequent power outages), faulty module and relative inexperience preparing samples.
External monitoring of error logs was useful for identifying common errors and methods to reduce error risk, and developed laboratory technicians’ skills. Fingerstick cartridges may reduce errors associated with correct sample preparation by removing the need for centrifuging and pipetting an exact amount of plasma.
GeneXpert maintenance and quality assurance
Use of GeneXpert for POC HCV VL testing required regular device maintenance and quality control measures for assays. The local distributor provided maintenance, replacing six Xpert modules and installing extra equipment twice; hence, our extended warranty was highly cost-effective. Tuberculosis testing studies using GeneXpert in many LMICs have reported Xpert module replacement rates of 32% within 11 months28 and 42% in 19 months.29 Our study replaced more modules than those cited above (to date, no HCV studies have reported this outcome); most were replaced at the MLF site, which suffered excessive dust and moisture.
Programme implementation studies of Xpert for tuberculosis on Mycobacterium tuberculosis complex (MTB) and resistance to rifampin (RIF) have highlighted that while ambient temperatures in rooms where Xpert is installed are important, many temperature-related errors are associated with inadequate exhaust of warm air from devices, either due to fan filters clogging with dust or inadequate air flow.28 29 Timely module replacement ensured rapid return to full capacity.30
The EQAS programme was difficult to establish because project staff were required to liaise with the study sponsor and Australia’s National Reference Laboratory (NRL) regarding enrolment processes, acquiring test samples and organising appropriate training, which was time-consuming. However once implemented, the programme ran smoothly. The cost of EQAS may be prohibitive for small programmes, unless integrated through national health laboratories in Myanmar. To date, no HCV implementation studies in LMICs report enrolment in EQAS programmes for community-based sites; it is unclear if this is due to non-reporting of operational components or non-implementation. Many implementation studies conducted globally are supported by Australia’s NRL, but are often short term and do not describe this aspect of implementation.
Stock storage
Storing stock at temperatures consistently under 30°C was difficult, partly due to high temperatures in Myanmar necessitating refrigerators and/or air-conditioned rooms. Overall, stock was managed adequately in this study, close to real-world implementation. Many staff at one site did not prioritise stringent stock storage. Sometimes project staff moved stock and troubleshooted barriers to correct storage with staff during site monitoring visits. Storing stock correctly is problematic in many community-based settings, especially those with high average temperatures and unstable electricity supply, as outlined in the Xpert MTB/RIF tuberculosis testing manual.31 To date, few studies of HCV VL implementation have addressed these operational challenges, which are critical for decentralised laboratories when implementing similar models of care.
HCV treatment workflow
Appointment scheduling and patient follow-up
Flexibility in appointment scheduling improved the model’s convenience for many patients living outside of Yangon and/or who needed methadone dosing every morning. A structured patient reminder and follow-up process aided high retention in care, but added to GPs’ and nurses’ workloads. Task-shifting from nurse/GP to reception or peer worker staff might reduce key clinical staff workloads, and automated SMS reminders could reduce time spent calling patients. There is strong evidence that reminders increase appointment attendance,32 but the most effective mode of delivery (eg, SMS, telephone call, letter) may depend on the patient group and healthcare service.33 Highly motivated, altruistic, well-educated and generous staff facilitated the care model and enabled intensive follow-up procedures.
Managing workflow and workload
Recruitment targets reflecting realistic patient flow were useful to manage clinical staff workload. For one GP provider, with nurse and laboratory technician assisting, a recruitment rate of eight patients per day (in addition to follow-up appointments) was achievable. In other settings, campaign-style recruitment can process substantially more participants. For example, in Egypt, 3 physicians, 7 nurses and 4 laboratory specialists screened 3663 people for hepatitis C and hepatitis B over 4 days.34
Note that staff at both sites were almost exclusively focused on providing HCV care to patients, whereas many GPs must balance various patient needs. GPs involved in delivering care in this study had treated HCV with DAAs in other research and philanthropic work. All clinical staff were committed to offering their patients a cure and to the study’s success, and had access to a broader implementation team to support model implementation and troubleshoot problems.
Workload could be reduced by eliminating monthly DAA dispensing and monitoring visits (ie, weeks 4, 8 and 12). This would also reduce patients’ transport costs and time off work, improving the model’s affordability and cost-effectiveness. Evidence suggests a minimal monitoring approach is safe, efficient and effective. The recent minimal monitoring (MINMON) approach trial tested having the entire 12-week course dispensed at the treatment initiation visit, no scheduled visits or laboratory monitoring, and only two points of remote contact scheduled to check adherence at week 4 and then to schedule SVR12 assessment appointment.35 Throughout, patients were able to attend for unplanned visits whenever they required assistance. The SVR rates were excellent, at 95% (379/399) for all those initiated treatment. No serious adverse events related to treatment were reported.35 Further work assessing the safety and effectiveness of this model for diverse patient groups may be required, for example, only 14% of the patient group reported current substance use; however, these results suggest no issues with cure rates using this minimal monitoring method.35
Patient education
General patient education covering pre-test and post-test counselling and DAA education was time-consuming, and providers felt that many patients did not absorb key information. Task-shifting to nurses or trained community health workers is common for patient education in HIV and maternal and child health programmes globally, including in LMICs,36–38 and could be explored for HCV.
Misconceptions about HCV transmission—including that HCV is transmitted through saliva or acquired through eating certain foods—appear to be common in Myanmar and hamper efforts to educate patients on avoiding reinfection. This is not confined to Myanmar; in China, mosquitoes and sharing food or utensils are commonly believed to transmit HCV.39
Sample transportation and results communication
While this model used rapid PoC testing for HCV diagnosis, timely blood sample transportation for positive patients from the decentralised clinic site to an external laboratory for processing was essential, facilitating timely treatment initiation and minimising loss to follow-up.
Referral pathway
The main operational challenges for GP-delivered care were developing clear referral criteria, an accessible referral pathway and GP-specialist communication. These challenges are not unique to this feasibility study; difficulties in deciding when to refer, navigating referral processes and achieving effective GP-specialist communication are common in many specialty areas.40 Our criteria for hepatologist review were stricter than the national guidelines require. They were intended to provide guidance and reassurance to treating GPs regarding hepatic decompensation assessment, and reduce the risk of patients being treated with DAAs instead of being evaluated by specialists as per the national guidelines. Understanding when and how to involve specialists in GP-delivered care is critical to safe and effective delivery of HCV care in the community.
Myanmar’s healthcare system restricts development of accessible referral pathways for specialist care. Public hospital outpatient departments do not offer scheduled appointments, but require patients to register and wait until called. Other options include referral to a rotating specialist at a non-governmental organisation (NGO) clinic, as in our study, or referral to private specialist clinics, which we did not promote due to them being prohibitively expensive for most patients. In addition, remote consultations could occur using video conferencing technologies or secure, group instant messaging services to review patients’ specific treatment plan remotely or to review as case presentations on a regular basis for ongoing education. Uptake of remote consultations has been spurred by the COVID-19 pandemic necessitating fewer face-to-face consultations.
Efficient referral and review processes are common challenges in healthcare, with various interventions trialled and evaluated.41 Our hub-and-spoke referral system was feasible, but could be improved via more options for referral to private clinics and financial support for this to facilitate timely review, and more direct GP-specialist communication to discuss cases such as through remote consultations.
Use of an electronic medical records system
Standalone training for use of OpenMRS for data collection was needed. Training was necessary to upskill staff to capture study data in this system, because many staff had never used electronic data capture tools and had minimal computer literacy.
Using OpenMRS improved data quality and was a good introduction to electronic data capture. It was not used for referral or sharing patient information with specialists, or for external laboratories to enter test results, or for real-time reporting to government or other bodies. However, these are technically feasible and logical applications.
Identifying capacity gaps and reducing identified constraints to scaling up
Assessing intervention scalability and preparing intervention scale-up plans requires considerable engagement with key stakeholders and resources. Generally, this involves structured thinking and decision-making guided by decision tools or frameworks.42 43 An intervention scalability assessment and a scale-up plan was outside the scope of this work; however, we provide some observations for those interested in replicating this model of care.
Implementation context is critical to intervention feasibility and success. Note that the staff involved here were invested in providing an accessible service to their patients and in the study’s success. The staff’s dedication to the service likely contributed to their acceptance of these challenges when providing a clinical service and their ability to adapt the service to overcome these challenges; the willingness of teams in other settings to overcome these challenges and adapt service delivery is unknown and should be considered when implementing in new contexts. Moreover, note that we covered the costs of all diagnostic tests, pretreatment assessments and treatment for patients (ie, patients paid nothing), which is unlikely to occur in a national government-funded primary care-based HCV programme. Paying for tests and treatment may reduce uptake of care drastically, and is important to evaluate to inform scale-up.
Briefly, scale-up of this model of care in Myanmar would require:
identification of appropriate decentralised sites, including harm reduction sites (such as needle/syringe distribution programmes, opioid substitution therapy dispensing sites), primary care clinics and HIV treatment sites;
an appropriately trained workforce, funding for human resourcing and development of referral pathways, given the very small hepatologist workforce in Myanmar;
a robust supply chain for DAAs and Xpert cartridges (noting that COVID-19 and political instability can disrupt previously well-established mechanisms); this could be managed centrally by the National Hepatitis Control Programme;
funding for diagnostics and DAAs, in a setting where even minimal out-of-pocket costs are unaffordable to most people;
promotion of the benefits and availability of HCV treatment to ensure take-up beyond the ‘willing and waiting’.
Scale-up of similar models of care within the primary healthcare system in Myanmar must be driven and supported by the Ministry of Health and Sports, in particular the National Hepatitis Control Programme.
Strengths and limitations of this study
The model being partially embedded in the healthcare system was a strength, in that patients were referred to public hospitals if required, the service settings were similar to those of other primary healthcare clinics and it allowed identification of specific scalability challenges. Another strength was the use of various data sources for evaluation, and high-quality clinical and operational data supported the feasibility evaluation and recommendations. While we had good quality data, we were unable to definitively determine the root cause of the higher GeneXpert error rate; detailed reporting of GeneXpert error rates and module replacement rates should be prioritised from other community-based HCV programmes in LMICs.
This model of care study’s external funding covered key implementation challenges, including refining laboratory set-up; lack of such support will threaten success in other settings. In addition, we evaluated the clinical model of care within a research setting, limiting its real-world applicability. Also, it should be noted that the staff at our clinics provided HCV testing and treatment as their primary work, as opposed to providing a broad suite of primary care services with HCV as one component. Also, for the purpose of this study, the clinics only treated those with HCV mono-infection; further modifications may be required to provide care to those with HIV or hepatitis B co-infection. The staff involved in providing clinical care were aware they were participating in a trial. This probably meant they invested more personal time and adjusted their standard of care to meet patients’ needs (within the protocol’s constraints); for example, being extra flexible with appointment times, calling patients to remind them of appointments, and occasional fundraising to cover the accommodation costs of patients from outside Yangon.
Finally, the study coordinator conducted provider interviews and, with the study manager, wrote the evaluation, feasibility assessment and scalability recommendations. Their investment in the trial’s success potentially biased the questions asked when interviewing and the responses of staff regarding challenges in study implementation, as well as the interpretation of data.
Conclusion
Implementing a decentralised HCV care model in urban Myanmar is feasible, and was very successful in retaining people in care through to cure, with high cure rates over 90% achieved across both sites. However, implementation in these challenging circumstances required service adaptations and ongoing dedication of our resilient staff to provide a successful care model. Staff were well supported by a broader study team, and GPs had access to ongoing specialist mentoring and support. Our evaluation of the feasibility of the intervention, and experiences implementing the project lead us to 10 key recommendations, including importance for continuous error monitoring, extended warranties, infrastructure improvements to enable appropriate stock storage and laboratory conditions and options for further task-shifting. Requirements for implementation and feasibility considerations identified herein will assist in adapting this model to other settings, including general primary care facilities, decentralised national HIV programme sites and other NGO-led sites, in Myanmar and globally. Growing evidence for the effectiveness of decentralised, non-specialist-led HCV care supports expansion of this type of care model. Continued research is needed to generate more evidence on how to implement and adapt to context to improve HCV models of care.
Data availability statement
No data are available. No unpublished data available for sharing.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by The Myanmar Department of Medical Research Ethics Review Committee (#2019-144) and The Alfred Hospital Human Research Ethics Committee (#244/17). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
Our study was implemented from January 2019 to August 2020; with all work completed prior to 1 February 2021. The authors acknowledge the collaboration and contributions of the National Hepatitis Control Programme of the Myanmar Ministry of Health and Sports, the Myanmar Liver Foundation, the study participants and the study team.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
BLD and WLY are joint first authors.
AP and MH are joint senior authors.
Contributors BLD and WLY led the preparation of the original draft, and subsequent editing, and had access to all the data. SS, JM, NN, HH, BLD and WLY led project administration of this work and contributed to the investigation processes. MH was responsible for funding acquisition. BLD, WLY, AB and AEP led data analysis and interpretation. PE, KPK, WN, KSA and TMW were involved in reviewing and editing the manuscript. MH, AEP and JH assisted with conceptualisation of the study, provided supervision, methodology support and reviewing and editing. BLD, WLY and MH accept full responsibility for the work and conduct of the study, and decision to publish. BLD and WLY are responsible for the overall content of the study and act as guarantors.
Funding The main study was supported by Unitaid, through the FIND-led HEAD-Start. Ancillary qualitative work was supported by Gilead, through Dr Win Lei Yee’s Public Health Award Fellowship (#05378). The Australian National Health and Medical Research Council (NHMRC) provided funding to AB through an Early Career Fellowship, to BLD through an NHMRC postgraduate scholarship, and to MH through an investigator fellowship. The authors also wish to acknowledge operational support through the Victorian Operational Infrastructure Support Program received by the Burnet Institute.
Disclaimer Funding bodies had no influence on interpretation of the findings, writing the manuscript or the decision to publish.
Competing interests MH has received investigator-initiated grant funding from Gilead Sciences and AbbVie for unrelated work. AP has received investigator-initiated grant funding from Gilead Sciences, MSD and AbbVie and speaker fees from Gilead Sciences for unrelated work. JH has received investigator-initiated grant funding and speaker fees from Gilead Sciences for unrelated work. WLY has received Gilead Sciences Fellowship for related work. KPK has received non-financial support from Mylan, Hetero and Royal Ruby. WN has received non-financial support from Mylan and Cipla. All others declare no potential competing interests.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.