Study aims
As above, we will conduct a type 2 hybrid effectiveness-implementation trial to evaluate an integrated NCD care management intervention in rural Nepal. The intervention is described in depth in the Supplementary File 1.
Study implementers
Healthcare workers and research staff from the non-profit organization Nyaya Health Nepal, their collaborators in the Ministry of Health and Population and Nepal Health Research Council, and collaborating researchers form the study team will lead the study. Nyaya Health Nepal has been working in a public-private partnership with the Ministry of Health and Population for over ten years in rural Nepal to deliver community- and facility-based health services, and this study will leverage this pre-existing partnership and care delivery network. Nyaya Health Nepal operates with a United States-based non-profit organization, Possible, to advance national and global healthcare systems policy and practice priorities.
Study setting
The study will take place in Achham and Dolakha districts of Nepal across four municipalities. Following recent healthcare decentralization, Nepal’s 750 municipalities manage primary healthcare delivery. The intervention will be implemented in a step-wise fashion in coordination with municipal-level government authorities and study staff.
Achham is a remote, impoverished district of 260,000 people, with large migrant populations and a history of social disruption during the Nepali civil conflict.[70-74] Achham has one of the highest district-level under-five mortality rates[75] and one of the lowest human development indices in the country.[76] The study implementers have been delivering some NCD-related care at the district-level Bayalpata Hospital and to communities within the hospital’s catchment population since 2008. Bayalpata Hospital serves approximately 90,000 outpatient and 3,000 inpatient visits per year. CHW services include proactive case detection, care coordination, and counseling. The study will include a catchment population of approximately 50,000 in Achham across two municipalities.
The second district is Dolakha, one of the hardest hit districts in the 2015 earthquakes.[77] Nyaya Health Nepal’s work in Dolakha is based at Charikot Primary Health Care Center, which serves approximately 60,000 outpatients per year, with similar CHW services to those in Achham’s. The study will include a population of approximately 30,000 in Dolakha across two municipalities. Thus, the total expected study population will be 80,000.
Within the context of the public private partnership between the government and Nyaya Health Nepal, no user fees are charged for any facility-based or community-based services, in either Achham or Dolakha, thereby mitigating financial access barriers to care delivery and study participation.
Within the study setting, MLPs for the NCD intervention are locally defined as the Nepali cadre of Health Assistants, who have three years of post-secondary medical training. The CHWs in this intervention have secondary-school-level education, and are fully employed, with on-going supervision by Community Health Nurses (CHNs). They receive initial training of approximately one month when they are hired, and on-going weekly trainings to continually improve their skillsets. The CHWs are employed by the public-private partnership between Nyaya Health Nepal and the Ministry of Health and Population. They are distinct from the robust Female Community Health Volunteer network that exists throughout Nepal,[63, 78] who have historically focused on vaccination, public health messaging, and other community preventive interventions rather than on household care delivery and follow-up. These staffing, supervision and training structures are described in greater detail in the Supplemental File 1 and 2.
Study populations
For primary quantitative outcomes, the study population will include adult patients (≥18 years of age) who qualify for a diagnosis of hypertension, type II diabetes, and/or COPD, according to WHO-PEN guidelines, and are engaged in longitudinal care by Nyaya Health Nepal’s team in Achham and Dolakha. The study will limit enrollment to the catchment areas served by both the facility-level and CHW-level services deployed by Nyaya Health Nepal. Study participants will be initially enrolled during facility-based visits at Bayalpata Hospital and Charikot Primary Health Care Center prior to the completion of intervention roll-out, and are considered engaged in longitudinal care if they have at least one follow-up hospital visit after 12 months of their initial visit. Digital health records that link between the facility-based EHR and the CHWs’ mobile-phone applications will be utilized to share patient data across settings, when available. CHWs can identify potential patients in the community and refer them to the facility for diagnosis confirmation, following which they could be included in the study. Patients’ receipt of care will not be contingent upon their enrollment in the study; all patients will continue to receive care per routine service delivery. This represents an exhaustive convenience sampling method as all eligible patients identified at Bayalpata Hospital and Charikot Primary Health Care Center may be enrolled in the study. Exclusion criteria are (1) individuals planning to migrate from the study area prior to twelve months of exposure to the intervention, or (2) individuals explicitly requesting exclusion from the study or declining to consent (see Supplemental File 3) for the study.
For implementation components, staff members, patients, community leaders, and government officials will be approached for key informant interviews (KIIs) and focus group discussions (FGDs), as described below.
Study design
This is a prospective, mixed-methods type 2 hybrid effectiveness-implementation study to evaluate an integrated NCD care management intervention. We plan to apply both qualitative and quantitative methods in a complementary manner,[79] in order to meaningfully assess both patient-level and population-level outcomes and the effectiveness of the implementation strategy. We will study the intervention’s impact on patients’ disease management outcomes after 12 months of being enrolled in NCD care using a pre-post design across both sites.
1) We will study the implementation of the intervention utilizing both quantitative and qualitative methods applying the RE-AIM (Reach, Efficacy, Adoption, Implementation, and Maintenance) framework [80].
Data collection is developed and integrated within the routine course of delivering care, which is an ethical, acceptable, and affordable approach in this setting. See Figure 1 for a trial flowchart and Supplemental File 4 for a SPIRIT research reporting checklist. is not feasible nor ethically acceptable to obtain data on a comparison (control) group in this population. Given the lack of national or local NCD systems, no data are available from other sources prior to the start of the study.
Study outcomes
The study has two Specific Aims: effectiveness (Specific Aim 1) and implementation (Specific Aim 2), as detailed in Table 1. For Specific Aim 1, the primary outcome will be the proportion of patients who meet disease-specific, evidence-based control measures at the completion of their initial twelve months engaged in treatment. These “at-goal” metrics aim to serve as simplified measures to assess disease control status, recognizing the limitations associated with multiple disease-specific metrics in settings like rural Nepal, especially for patients with multiple co-morbid conditions. These are presented in Table 2.
Table 1: Metrics for Specific Aim 1 (efficacy) and Specific Aim 2 (implementation)
Aim
|
Outcome/ RE-AIM Element
|
Indicator
|
Definition
|
Specific Aim 1: Efficacy
|
Primary outcome: control of NCD conditions
|
Condition-specific “at goal” metrics
|
-% of enrolled NCD patients achieving “at goal” status (Table 2), at the completion of the study period
|
Secondary outcome 1: tobacco use
|
Tobacco use status
|
-% of enrolled NCD patients who were using tobacco at enrollment who are non-users or who have reduced by >50% their tobacco intake, at the completion of the study period
|
Secondary outcome 2: alcohol use
|
Alcohol use status
|
-% of enrolled NCD patients who were alcohol drinkers at enrollment who are non-drinkers or who have reduced by >50% alcohol intake, at the completion of the study period
|
Specific Aim 2: Implementation
|
Reach
|
Home visit coverage
|
-% of enrolled NCD patients having a CHW home visit, measured monthly
|
Clinic visit coverage
|
-% of enrolled NCD patients having an MLP visit at the clinic, measured monthly according to the patients indicated to be seen that month based on protocol-based guidelines
|
Demographic, geographic barriers and facilitators
|
-% of enrolled NCD patients whose CHW has GPS-mapped their households, describing barriers/facilitators to individuals’ access, and identifying contributors to variation/inequities
|
Loss to follow-up
|
-% of patients, stratified by demographic data and NCD conditions, that are lost-to-follow-up after enrollment
|
Monthly patient touch-points
|
-Number of monthly per-patient touch-points, including interactions by both MLPs and CHWs
|
Efficacy
|
Evidence-based hypertension management
|
-% of enrolled hypertension patients in accordance with evidence-based recommendations, as prescribed by clinical algorithms, assessed quarterly by EHR audits
|
Evidence-based diabetes management
|
-% of enrolled diabetes patients in accordance with evidence-based recommendations, as prescribed by clinical algorithms, assessed quarterly by EHR audits
|
Evidence-based COPD management
|
-% of enrolled COPD patients in accordance with evidence-based recommendations, as prescribed by clinical algorithms, assessed quarterly by EHR audits
|
Adoption
|
Village cluster adoption
|
-% intended village clusters receiving intervention
|
Timely adoption
|
-% intended village clusters rolling-out intervention within 3 months of schedule, according to local governance decisions to roll-out the intervention
|
CHW adoption
|
-% CHWs trained in intervention implementation within first six months
-% of trained CHWs retained in their positions at the completion of the study period
|
MLP adoption
|
-% MLPs trained in intervention implementation
-% of trained MLPs retained in their positions at the completion of the study period
|
Implementation
|
Care integration
|
-% of all NCD patients enrolled at the facilities seen by CHWs at home within first month
|
CHW supervision model
|
-% scheduled CHW supervision field visits completed, stratified by CHN and district, measured quarterly
-% of scheduled quarterly data review meetings held with CHWs and CHNs, measured quarterly
|
CHW home visit fidelity
|
-% of enrolled NCD patients with 100% of algorithm-indicated home visits received
-% of topics included at each session as dictated by the condition-specific algorithms, assessed during the CHW supervision field visits by CHNs, measured quarterly
|
Referrals
|
-% of patients appropriately referred to MLP care as indicated by the clinical algorithms, assessed during the CHW supervision field visits by CHNs, measured quarterly
-% of patients referred by CHWs seen by MLPs within the prescribed time window according to the clinical algorithms (e.g. 24hours, 72 hours, 1 week), measured quarterly
|
MLP supervision model
|
-% of enrolled NCD patients appropriately referred to see a physician by MLPs as indicated by the clinical algorithms, assessed during monthly physician supervision sessions, measured quarterly
|
MLP visit fidelity
|
-% of enrolled NCD patients with 100% of algorithm-indicated facility visits received, assessed during monthly physician supervision sessions, measured quarterly
-% of diagnostic, treatment, and counseling topics included at each session as dictated by the condition-specific algorithms, assessed during monthly physician supervision sessions, measured quarterly
|
Implementation challenges
|
-Exploratory and hypothesis-generating as revealed through FGDs and KIIs with CHWs, CHNs, MLPs, physicians, patients, and other relevant community stakeholders
|
Maintenance
|
Total intervention cost
|
-Cost of each intervention component and total costs using the Joint Learning Network costing methodology
|
Intervention initiation costs
|
-%breakdown of initial (one-time) costs for intervention (training, equipment, etc)
|
Intervention maintenance costs
|
-% breakdown of maintenance (recurring) costs (on-going training, personnel, materials, and other)
|
Facility vs. community costs
|
-% of costs of health care divided between facility level and community level
|
Geographic cost variation
|
-Characterization of variance in costs between village clusters and districts within the intervention catchment area
|
Out-of-pocket patient costs
|
-% costs of health care divided between facility level and community level
|
Integrated intervention cost-effectiveness analysis
|
-Pre/post intervention marginal effectiveness for primary outcomes
|
|
|
Cost per unit
|
-Intervention cost per enrolled patient
-Intervention cost per capita
-Projected cost to scale intervention nationally, based on known incidence and prevalence of each condition
|
Table 2: Clinical definitions of “at-goal” status for each intervention condition
Non-communicable disease
|
Management metric
|
“At-goal” definition
|
Type II diabetes mellitus
|
Hemoglobin A1c OR fasting blood sugar
|
Hemoglobin A1c < 7.5 OR fasting blood sugar <130 mg/dL*
|
Hypertension
|
Blood pressure
|
Blood pressure <130/80mm Hg or patient-tailored goal per risk stratification^
|
Chronic obstructive pulmonary disease
|
Exacerbation status
|
≤1/3 Anthonisen criteria ¥
|
Footnotes:
*Type II diabetes mellitus:
The 2018 American Diabetes Association guidelines [81] call for a goal A1c <7% for most patients or A1c<8% in "patients with a history of severe hypoglycemia, limited life expectancy, advanced microvascular or macrovascular complications, extensive comorbid conditions, or long-standing diabetes in whom the goal is difficult to achieve despite diabetes self-management education, appropriate glucose monitoring, and effective doses of multiple glucose-lowering agents including insulin." For our intervention, we established 7.5% as our goal to pragmatically accommodate both populations.
^Hypertension:
Based on the 2017 American College of Cardiology and American Heart Association guidelines,[82] we established <130/80mm Hg as a default goal, with patient-tailored goals for select patients (≥65 years of age, multiple co-morbidities, limited life expectancy, clinical judgement, patient preference).
¥Chronic obstructive pulmonary disease:
The 2017 update to the GOLD guidelines [83] define COPD exacerbation as an "acute worsening of respiratory symptoms that results in additional therapy." We used the Anthonisen criteria of worsening sputum volume, sputum purulence, and increased dyspnea to define the “worsening of respiratory symptoms” specified in the GOLD guidelines. We established a threshold of no more than one Anthonisen criterion as a pragmatic tool for determining clinical status.
Secondary outcomes for Specific Aim 1 will include the following. We will assess the individual “at goal” rates per condition. We will assess the persistence of the intervention for the subset of patients for whom we have the data (i.e., those enrolled within 12 months of the start of the study) on their 24 months’ outcome. Additionally, we will examine the tobacco and alcohol status of enrolled patients, specifically focusing on the proportion of patients who were tobacco users and/or alcohol drinkers at the time of enrollment, who have stopped all tobacco and / or alcohol intake, or reduced their intake by >50%, by the completion of the study period (Table 1).
For Specific Aim 2, the RE-AIM framework will be utilized to assess the implementation of the study intervention, with RE-AIM metrics as listed in Table 1.
Sampling strategy and power calculations
We will use exhaustive convenience sampling to screen all eligible patients seen across two facilities over a twelve month period into the analysis cohort. Based on historical formative data of patient volume seen at these two facilities, and accounting for an expected 30% attrition rate, we conservatively expect that at least 1000 patients will be eligible for enrollment into the cohort.
With this conservative number of 1000 as our expected sample size based on this convenience sampling, we conducted power calculations to determine the statistical power to detect a change in the “at goal” status. We calculate power based on a simplified design to compare paired proportions using a two-sided McNemar’s test with an 0.05 Type I error (alpha) level. The primary outcome is the proportion of patients who achieve their NCD control target (“at goal status”) after 12 months of being engaged in care. We used SAS version 9.4 (Cary, NC) to estimate power to detect a 5% difference between discordant proportions, i.e. proportions of patients whose “at goal” status changed from “not at goal” at baseline to “at goal” at follow-up, and vice-versa, in multiple scenarios where the total proportion of discordant patients ranged from 10% to 40% of all patients. Based on these assumptions, our power to detect a 5% difference in the discordant pairs is 69%, when the total discordant proportion is 40%, and the power is 99% when 10% of all patients were discordant.
Data collection
Quantitative data
Quantitative data for patient outcomes will be extracted from the facility-based EHR and the CHW’s mobile phone application (Supplemental File 5), and will be used to assess Specific Aims 1 and 2. (Table 1) All implementation-related data for evaluating the performance of MLPs and CHWs (Table 1) will be collected by the responsible MLP and CHW supervisors in digitized checklists within the EHR and mobile phone application. Access to protected health information will be controlled and defined by user access groups according to clinician status. Data to be analyzed will be extracted via secure data queries from the EHR system in aggregate, partially de-identified form with external researchers signing a data sharing and use agreement prior to analysis. Cleaned, de-identified datasets will be made publicly available via a data repository.
Costing data for the intervention will be collected utilizing a “top-down” method, as described by the Joint Learning Network[84]. This method will document direct and indirect costs associated with the NCD care delivery intervention described here and related administrative functions (including planning and administration; training; supervision and monitoring and evaluation; data management; and continuous surveillance) will be disaggregated. Full methodology of direct and indirect costs is provided by the Joint Learning Network[84], and will be utilized for this study. For the purposes of this pragmatic study, this methodology will be appropriate to estimate the additional marginal costs of the intervention (rather than cost-savings or secondary cost implications) as compared to general standard of care.
Qualitative data
Qualitative data will be used for Specific Aim 2. (Table 2) Staff members, patients, community leaders, and government officials will be approached for KIIs and FGDs. Purposive sampling will be used, aiming to maximize heterogeneity across sex, socioeconomic position, healthcare issues, geographic location, age, caste-class, and other attributes. For each group, five key informant interviews will be conducted at each time point, as described below. One focus group discussion per group will be conducted at each time point.
KII and FGD guides will be developed in advance, and will vary across the study period, exploring specific topics of concern. A locally validated, seven-domain framework of healthcare delivery analysis will be used to inform data collection.[85] These seven domains include health service operations, supply chains, equipment, personnel, outreach, societal factors, and structural factors. Qualitative data collection will focus on these areas to assess the implementation of the intervention.
FGDs and KIIs will occur prior to the initiation of the intervention, and in intervals of six months throughout the study period, to assess on-going implementation status. All sessions will be conducted in Nepali. All qualitative data will be stored on a Research Electronic Data Capture (REDCap) database.[86] REDCap user access will be defined so that researchers only have access to de-identified study data. Any paper copies of data forms will be stored in locked cabinets inside locked rooms at district facilities. Once all data are fully transcribed and validated for quality, all paper copies will be destroyed. REDCap data will be deleted twelve months after the completion of the study period.
Data analysis
Analysis for Specific Aim 1: effectiveness
In order to assess the effectiveness of the intervention, as described above, the primary outcome will utilize disease-specific “at goal” metrics for each of the three study diseases: hypertension, type II diabetes, and COPD. We hypothesize that the integrated intervention will lead to a 10% increase in the “at goal” status of the combined disease cohorts, over a twelve-month follow up period.
We will use conditional multivariable logistic regression to assess patient outcomes at 12 months follow up, adjusting for potential confounding and/or effect modification by patients’ demographics (including age, sex, caste), municipality, district, mean distance to the hospital, and engagement in care (defined as number of facility-based and community-based encounters). We additionally hypothesize a 10% improvement in the status of each of the two secondary outcomes: tobacco and alcohol use, as measured by patient-reported outcomes in Table 1.
As a secondary analysis for Specific Aim 1, namely the time-varying nature of the outcomes, we will assess the longitudinal effect of the intervention, as measured in three-monthly intervals, throughout the study period, compared to baseline statistics at the time of each village-cluster enrollment. Variables will be considered as either nominal or continuous (linear effect) predictors, and the generalized linear model framework will be used to estimate effect of time-varying repeated measure intervention implementation over the several steps of the wedged design. Differential impact from time of intervention will be evaluated with test of month × intervention interaction. Models will be fit using generalized estimating equations, e.g., using SAS Proc Genmod, to calculate valid standard errors in the presence of repeated measures over time and possibly correlated outcomes at the municipality level. Assumptions of over- or under-dispersion will be examined closely, and an estimated scale parameter or negative binomial models will be used as needed.
Analysis for Specific Aim 2: RE-AIM implementation framework
In this mixed-methods study, Specific Aim 2 will be assessed using the RE-AIM framework for implementation trials.[80] A full list of metrics, separated by each domain of the RE-AIM framework, is presented in Table 1. Additional details regarding the supervision and audit structure for MLPs and CHWs can be found in Supplemental File 1 and 2.
For the (M)aintenance of the intervention, we will assess the costs of the intervention, using the Joint Learning Network methodology[84]. Cost data will be analyzed and presented (Table 1) to help program planners and policy makers understand the implications for possible scale of a similar intervention by the government or other entity in the future.
For quantitative data within Specific Aim 2, a similar methodology of generalized estimating equations, as described above in the section on Specific Aim 1 analysis, will be applied. Data will be assessed in three-month intervals.
For qualitative data within Specific Aim 2, analysis will be on-going and iterative, so as to continually inform further qualitative data collection, focusing on timely and relevant implementation issues. Data from KIIs and FGDs will transcribed and coded using Grounded Theory Methodology.[87, 88] NVivo software will be used for qualitative data analysis.[89]