The authors have declared that no competing interests exist.
The burden of chronic, non-communicable diseases such as diabetes is growing rapidly in low- and middle-income countries. Implementing management programs for diabetes and other chronic diseases for underserved populations is thus a critical global health priority. However, there is a notable dearth of shared programmatic and outcomes data from diabetes treatment programs in these settings.
We describe our experiences as a non-governmental organization designing and implementing a type 2 diabetes program serving Maya indigenous people in rural Guatemala. We detail the practical challenges and solutions we have developed to build and sustain diabetes programming in this setting.
We conduct a retrospective chart review from our electronic medical record to evaluate our program’s performance. We generate a cohort profile, assess cross-sectional indicators using a framework adapted from the literature, and report on clinical longitudinal outcomes.
A total of 142 patients were identified for the chart review. The cohort showed a decrease in hemoglobin A1C from a mean of 9.2% to 8.1% over an average of 2.1 years of follow-up (p <0.001). The proportions of patients meeting glycemic targets were 53% for hemoglobin A1C < 8% and 32% for the stricter target of hemoglobin A1C < 7%.
We first offer programmatic experiences to address a gap in resources relating to the practical issues of designing and implementing global diabetes management interventions. We then present clinical data suggesting that favorable diabetes outcomes can be attained in poor areas of rural Guatemala.
The global burden of chronic, non-communicable diseases (NCDs) like diabetes is growing rapidly [
This study focuses on diabetes in Guatemala, an LMIC in Latin America with a population of approximately 16 million people. The most robust data on the diabetes burden in Guatemala come from a 2006 survey showing an adjusted adult diabetes prevalence of 9.6%, which was similar to the U.S. prevalence when the study was conducted [
Given the epidemiologic data above, implementing high-quality clinical management programs for diabetes and other chronic diseases for populations in LMICs like Guatemala is a critical public health priority [
Many factors limit access to good diabetes care in LMICs [
This study reports on the design, implementation, and evaluation of an adult type 2 diabetes program serving indigenous populations in rural Guatemala. Our objectives are twofold. First, we share design and implementation experiences with our adult type 2 diabetes management program and, second, we conduct a chart review to assess our program’s clinical outcomes. The overarching goal is that this study is to assist fellow global health practitioners in implementing high-quality diabetes clinical programming tailored to other low-resource settings in LMICs.
The work in this study was conducted through our affiliation with Wuqu’ Kawoq | Maya Health Alliance (
The diabetes program operates in four central highland towns in rural Guatemala. These towns are predominantly Maya indigenous in terms of culture and language, possess poverty rates ranging from 60–85% [
We previously conducted formative research on patients with diabetes presenting to our primary care clinics [
The program provides ambulatory diabetes management and coordinates referrals for acute care and specialty medical appointments. All services are free, including reimbursement for transportation expenses. The program’s primary diabetes providers are bilingual (Mayan and Spanish) nurses. Nurse-directed care was implemented because physician density is very low in rural Guatemala [
Challenge | Opportunities | Solution |
---|---|---|
Cultural and linguistic barriers to biomedical care for indigenous Maya population | Rising number of educated young Maya professionals experienced in issues related to language, cultural, and health advocacy | Employ exclusively native speakers of Mayan languages as front-line health providers |
Low physician density | Large labor pool of indigenous nurses | Nurse-driven diabetes protocols |
High cost of diabetes medications | Dynamic generics industry in Guatemala | Limited formulary of locally purchased generic drugs |
Low availability of laboratory diagnostics | Validated point-of-care tests available on the local market | Use of point-of-care laboratory testing for hemoglobin A1C |
Patients live in rural, difficult-to-access villages | • Availability of open-source electronic medical record platforms |
Deployment of smart-phone-based data entry and open-source electronic medical record |
Low levels of education and health literacy | • Patients often work from home or return to home during day |
Home visit program by diabetes educator using locally adapted curriculum |
Patients present late in disease course, often with significant end-organ damage | Excellent subspecialty care available in capital city | Centralized case management system to coordinate referrals from rural health workers to pre-selected subspecialty clinics |
Chronic disease care is expensive and requires long-term commitments to beneficiaries | Blended financing models are emerging in global health | • Crowdfunding provides funding for extraordinary and catastrophic care |
Diabetes nurses manage glycemic treatment using a step-wise clinical algorithm developed from national [
New or uncontrolled patients are offered regular home education visits, which are conducted by a bilingual diabetes nurse educator using a curriculum adapted from the U.S. National Heart, Lung and Blood Institute’s
Our program formulary consists solely of generic medicines, including common oral diabetes drugs (metformin and glyburide), insulin (NPH and regular), and anti-hypertensive agents (ACE inhibitors, among others). Laboratory tests are carried out using point-of-care devices when feasible or in fee-for-service regional laboratories. We monitor glycemic control with point-of-care hemoglobin A1C testing; while this test is more expensive than blood glucose monitoring and not routinely used in rural Guatemala, we have found it to be an indispensible clinical tool.
Bilingual (Spanish-Mayan) caseworkers provide transportation, accompaniment, and interpretation for patients referred from rural communities to urban health facilities. Patients with acute diabetes-related complications are referred to national or regional public hospitals where caseworkers advocate for them. Patients with non-acute specialty medical needs are referred to public or private facilities with which we have developed working relationships.
Financial sustainability has been a central challenge given that diabetes patients require long-term commitments, per-patient costs rise over time as disease severity progresses, and the global funding landscape for diabetes programs is limited. Our core diabetes expenses are supported through donor fundraising. We partner with a popular global health crowdfunding platform, Watsi (
In our retrospective chart review, we identified active adult type 2 diabetes patients who had been enrolled in the diabetes program for at least 6 months as of 1 May 2015. We defined an “active patient” as having had at least one clinical encounter documented in the one-year period from 5/1/2014-5/1/2015. The electronic medical record (EMR) utilized in our program is OpenMRS (
Our pre-defined search inclusion criteria included age ≥18 years and diabetes diagnosis as defined by a hemoglobin A1C ≥6.5, random blood glucose ≥200 mg/dL, history of a diabetes-related prescription (metformin, sulfonylurea, or insulin), or assignment to the diabetes module or problem list in the EMR. A total of 236 hits were initially generated. Each record was then manually reviewed to remove patients meeting exclusion criteria: residence in a non-program community, no visit documented in the defined time period, program enrollment < 6 months, diagnosis of type 1 or gestational diabetes, and erroneous EMR entries. A final list of 142 patients was identified. Data on these patients were manually extracted from the EMR to a spreadsheet, and separate authors reviewed the entire spreadsheet for errors. Variables extracted were demographic (date of birth, gender, preferred language, years of education, municipality of residence), historical (years with diabetes diagnosis, date of program enrollment), and clinical (hemoglobin A1C, systolic and diastolic blood pressures, height and weight, creatinine, proteinuria, medication prescriptions, frequency of encounters).
Chart review data was imported from a spreadsheet into Stata version 13 (College Station, TX) for statistical analysis. A demographic and clinical profile of the cohort was first generated using descriptive statistics. Subsequently, to assess cross-sectional outcomes, we adapted a framework for use of electronic heath records in evaluating quality of diabetes care in LMICs [
A cohort profile including demographics, history of diabetes and program enrollment, and basic clinical data is outlined in
Characteristic | Value |
---|---|
56.1 ± 11.8 | |
80.3 | |
Kaqchikel Mayan | 50.8 |
Spanish | 37.7 |
K’iche’ Mayan | 11.5 |
Grades completed, median (IQR) | 2 (0–4) |
Completed primary school–% | 20.8 |
Median (IQR) | 7 (4–12) |
47.2 ± 11.7 | |
Median (IQR) | 2.5 (1.3–3.8) |
For continuous variables with normal distribution, values are given as mean ± standard deviation. For continuous variables with nonnormal distribution, median and interquartile range (IQR) are specified. Some non-clinical data including preferred language, years with diabetes, and education attained were not available for all patients, as indicated by
Relevant clinical indicators, summarized in
Clinical Characteristic | Value |
---|---|
8.1 ± 2.1 | |
Diagnosis of hypertension–% | 45.8 |
Systolic BP, mean–mmHg | 121.8 ± 20.4 |
Diastolic BP, mean–mmHg | 74.9 ± 10.2 |
Mean | 28.0 ± 5.0 |
BMI ≥ 25 –% | 70.8 |
BMI ≥ 30 –% | 30.8 |
GFR 30–60 –% | 40.1 |
GFR ≤ 30 –% | 3.5 |
Proteinuria–% | 33.6 |
On dialysis–% | 2.1 |
Metformin | 85.9 |
Sulfonylurea | 44.4 |
Insulin NPH | 25.4 |
Insulin regular | 2.8 |
No insulin or oral anti-diabetic agent | 5.0 |
ACE inhibitor | 43.7 |
Median | 11.5 |
Interquartile range | 8–15 |
BP, blood pressure; BMI, body mass index, GFR, glomerular filtration rate.
a GFR was estimated from clinical variables using the CKD-EPI equation.
An overview of medication prescriptions revealed that metformin was the most commonly prescribed drug (85.9%), that approximately one-quarter of patients had been prescribed insulin, and that 44% of patients were prescribed an ACE inhibitor. Only 5.0% of patients were not prescribed any oral diabetes medicine or insulin. The median number of diabetes-related encounters per patient per year was 11.5 (IQR 8–15).
Cross-sectional indicators are displayed in
% | |
At least one measurement of hemoglobin A1C | 99.3 |
Comprehensive foot evaluation | 98.6 |
Measurement of creatinine and rate of glomerular filtration | 90.1 |
Four or more clinical encounters during year | 95.1 |
Diabetes self-care education provided in home visit | 50 |
Overweight/obese (BMI ≥25 kg/m2) patients with hemoglobin A1C ≥ 6.5 who received metformin, unless contraindicated (n = 66) | 98.8 |
Patients with hemoglobin A1C ≥ 8% who received insulin (n = 64) | 37.5 |
Patients with hypertension receiving inhibitors of angiotensin converting enzyme or angiotensin-receptor blocker, unless contraindicated (n = 65) | 95.4 |
Hemoglobin A1C < 8% in last measurement (n = 142) | 54.9 |
Blood pressure <140/90 mmHg in last 3 measurements (n = 139) | 59.0 |
Hemoglobin A1C < 8% in last measurement and blood pressure <140/90 mmHg in last 3 measurements (n = 139) | 29.5 |
In terms of health outcomes, 57% of the cohort met the program’s glycemic goal of hemoglobin A1C < 8% at their last recorded measurement, and 59% had well-controlled blood pressures at the most recent three measurements. Overall, 29.5% of patients met the composite indicator of both hemoglobin A1C < 8% in last measurement and blood pressure <140/90 mmHg in last 3 measurements. Of note, 32.4% of patients had a most recent hemoglobin A1C < 7%.
Characteristic (n = 142) | Initial | Last | p-value |
---|---|---|---|
Hemoglobin A1C –% | 9.2 ± 2.4 | 8.1 ± 2.1 | 0.00 |
Systolic BP–mmHg | 124.3 ± 20.0 | 121.8 ± 20.4 | 0.17 |
Diastolic BP–mmHg | 77.6 ± 11.2 | 74.9 ± 10.2 | 0.02 |
Percent of patients with hemoglobin A1C < 8% | 38.0 | 54.9 | 0.001 |
Percent of patients with blood pressure <140/90 mmHg | 69.7 | 73.2 | 0.46 |
Percent of patients with hemoglobin A1C <8% and blood pressure <140/90 mmHg | 25.4 | 38.0 | 0.01 |
BP, blood pressure.
We found that the mean hemoglobin A1C in the cohort showed a statistically significantly decrease from 9.2% ± 2.4 on initial presentation to 8.1% ± 2.1 at the most recent measurement (p = 0.00). Mean blood pressure also decreased, though the difference was only significant for mean diastolic blood pressure (p = 0.02) and not systolic blood pressure (p = 0.17). We also observed a statistically significant increase in the proportion of patients who exhibited adequate glycemic control (p = 0.001), as well as the percent of patients meeting the composite goal of A1C <8% and blood pressure <140/90 mmHg (p = 0.01). No significant change was found in the percent of patients meeting blood pressure goals alone.
This study describes the design, implementation, and outcomes of a type 2 diabetes program for Maya indigenous adults in rural Guatemala.
In the first part of the paper, we offer programmatic experiences to address a gap in published accounts of the practical issues of designing and implementing diabetes management interventions in LMICs. Although there is an increasingly robust literature on diabetes prevention, screening, education, and self-management in global health settings [
Several design and implementation experiences detailed in this paper may be generalizable to practitioners in other settings. First, in our program, nurses provide the bulk of direct patient care, education, and care coordination. In our context, we have found that there is a surplus of well-trained nurses who speak indigenous Maya languages, that well-supported nurses can provide high value diabetes care for their cost, and that nurses are motivated to carry out time-consuming, high-contact activities such as home visits, education sessions, and social work tasks. The model of “task-shifting” from to non-physician providers has been shown to be effective in other settings [
Our experience also reveals some of the practical challenges in adapting widely-accepted diabetes clinical standards to local environments where resources are limited and risk-reward trade-offs are unique. We have observed that the underlying assumptions of published international guidelines sometimes do not hold in our setting. An example of this is the use of hemoglobin A1C, a test that is recommended in all guidelines but is rarely used in rural Guatemala due to its expense and unavailability. Like Partners In Health’s experience in Rwanda [
Additionally, international diabetes guidelines typically recommend stricter glycemic targets of hemoglobin A1C ≤ 7.0. Not only does this recommendation assume that hemoglobin A1C testing can always be utilized, it also does not take into account the much greater risks of hypoglycemia in rural, isolated towns with limited emergency services compared to the high-income countries where the major diabetes clinical trials (such as the ACCORD [
Finally, we describe our blended diabetes financial model, consisting of operating funds from donations, crowdfunding, and grants. Financing is a primary challenge for our program given the limited funding environment for adult NCD services compared with infectious diseases and maternal and child health programs. Additionally, diabetes is an incurable disease that progressively worsens over time, and each patient we enroll requires a long-term financial commitment. Ultimately, providing high-quality diabetes care is expensive compared to other global health interventions. We attempt to keep costs down by utilizing non-physician (nurse) providers, restricting our formulary to generic drugs, and negotiating with our suppliers. A future research aim is to conduct a cost-effectiveness analysis to better elucidate the costs and benefits of our program.
In the second part of this paper, we conduct a retrospective chart review to examine program outcomes. In terms of clinical indicators, we found that the cohort showed a decrease in hemoglobin A1C from a mean of 9.2% to 8.1% over an average of 2.1 years of follow-up. The proportions of patients meeting glycemic targets were 57% for the less strict targets that we use programmatically (hemoglobin A1C < 8%) and 32% for the stricter threshold (hemoglobin A1C < 7%).
Our results appear favorable compared to other rural type 2 diabetes treatment programs in resource-limited settings. For example, a nurse-led program with 80 patients in Kwazulu Natal, South Africa found average hemoglobin A1C to be 10.8% at baseline, 8.4% at two years, and 9.7% at four years [
Despite clear difficulties comparing our small cohort with data from other global health sites and national-level data in the U.S., we offer preliminary results suggesting that good diabetes outcomes can be attained in poor areas of rural Guatemala with a cohort of patients who predominantly speak Mayan indigenous languages, have low levels of education, and have high rates of overweight and obesity. We hypothesize that important drivers of favorable clinical outcomes in our context arise from a design process that has emphasized frequent contact and collaboration with patients and their families as indicated by the high frequency of patient interactions (95% of with four or more clinical visits during the year, median 11.5 encounters per year), our use of highly competent and motivated indigenous nurses as primary diabetes providers, and the high proportion of patients receiving home education visits (50% of patients).
Areas meriting additional investigation and quality improvement initiatives include an examination of the factors contributing to the gender disproportion in our cohort (80% female), further inquiry into dietary aspects such as food insecurity that may contribute to the high observed prevalence of obesity (30.8% with BMI ≥ 30), and better understanding barriers to insulin utilization since 38% of patients with uncontrolled diabetes were not prescribed insulin. On this final point, there is a large literature primarily originating from high-income countries relating to the fear of insulin, or “psychological insulin resistance” [
This study has several weaknesses or limitations. First, our program and sample size is small and may not generalize to other institutional context or settings within Guatemala. Second, cost is a significant barrier to scaling-up diabetes programs in LMICs, yet we cannot at this time offer a robust cost- effectiveness analysis of our program. We are, however, able to use institutional balance sheets to estimate average cost/patient/year of $220. Third, retention is a critical aspect of diabetes treatment programs in global health, and it is especially important in settings like Guatemala where fragmented care is a fundamental feature of the experience of indigenous people with diabetes [
Current protocol used in diabetes programming for Wuqu’ Kawoq | Maya Health Alliance.
(PDF)
Home education manual adapted from U.S. National Heart, Lung and Blood Institute’s
(PDF)
We thank Daniel Tse, Chase Adam, and Grace Garey at Watsi for their assistance with and vision the crowd-funding model. We thank Merida Coj and Katia Cnop at Wuqu’ Kawoq for their assistance with care coordination. We thank Carol Riquiac Teleguario at Wuqu’ Kawoq for her tireless work in pioneering home-based diabetes education.