STRENGTHS AND LIMITATIONS OF THIS STUDY
Data comes from one health district and cannot be generalised to other countries in Latin America.
Use of mixed methods to provide possible explanations to observed changes in quantitative clinical outcomes.
Use of multiple imputation to investigate the impact of missing data.
The SARS-CoV-2 pandemic has caused large-scale social disruption, and the mortality and morbidity resulting directly from the virus is one of the worst health crises in our era. The pandemic has impacted the ability of health systems to maintain access to basic services such as preventative health, maternal child health and the management of non-transmissible chronic diseases.1–3
This study investigates change in access and delivery of health services for type 2 diabetes. Mortality and morbidity analyses have demonstrated that people living with type 2 diabetes have a higher risk of severe consequences when they contract COVID-19, and increasing levels of hyperglycaemia are correlated with worse clinical prognosis.4 5
Various studies have demonstrated the impact of the pandemic on the care of patients living with diabetes. Some studies with individuals living with type 1 diabetes have demonstrated improvements, possibly because restrictions on movement and social interactions reduced external factors that contributed to worse clinical control.6 However, the majority of studies have demonstrated an excess of morbidity and mortality for people living with type 2 diabetes during the pandemic, equally attributed to the increased relative risk after contracting COVID-19 as well as for the eventual interruptions in preventative and curative health services.7–10
Despite the fact that health systems in low-income and middle-income countries are more sensitive to the impact of the pandemic, the majority of research has been conducted in high-income countries.2 11 This gap in evidence is especially important in Latin America, a region with high levels of inequality and Indigenous populations living in a vulnerable state.12–14 A study in Ecuador demonstrated that during the pandemic, the rate of excess mortality for all causes was four times greater for Indigenous peoples.15 Likewise, in Guatemala, an excess rate of mortality was found, although it did not disaggregate by ethnicity.16
In order to further explore the impact of COVID-19 on the care of type 2 diabetes in vulnerable populations in Latin America, we carried out a mixed methods analysis of an intervention to expand a diabetes self-management education and support intervention in rural Indigenous communities in Guatemala. This intervention was in process when the pandemic started. Our quantitative data help to estimate the impact of the pandemic on an Indigenous community in terms of glycaemic control, risk factor management and quality of life. Complementary qualitative data from interviews with healthcare providers and individuals living with diabetes provide possible explanatory mechanisms for the observed quantitative changes.15 16
We used a parallel convergent mixed methods study design. We analysed quantitative data on diabetes control while at the same time conducting semistructured interviews with healthcare providers and individuals living with type 2 diabetes.17 We prepared this manuscript according to the Good Reporting of A Mixed Methods Study (online supplemental annex 1).18
This study was carried out by Wuqu’ Kawoq | Maya Health Alliance, the Institute of Nutrition of Central America and Panamá and the Inclusive Health Institute, in collaboration with authorities from the health district of Guineales and the Guatemalan Ministry of Health.
Study data comes from the Guineales health district, which consists of approximately 44 000 people located in the municipality of Santa Catarina Ixtahuacán in the department of Sololá. The majority of the population is Maya K’iche’ (94%) and work in agriculture. Public health services in the district include a main health post, 19 decentralised clinics and 74 healthcare providers.
The Guineales district was one of nine participating districts in an implementation study to scale up a type 2 diabetes self-management education and support intervention in Indigenous areas of Guatemala. The intervention implementation processes and effectiveness results have been published for the other eight districts.19 Significant improvements in glycaemic control, blood pressure and various self-care indicators were found. However, these eight other sites began the study much earlier and were in the process of finishing their respective intervention when COVID-19 movement restrictions were implemented in Guatemala. In Guineales, the majority of participants had just started the intervention when interrupted by the pandemic. However, we had collected baseline data and, when the pandemic restrictions were lifted, we returned to repeat this evaluation. We use this data to explore the impact of the pandemic on the care of diabetes in Guineales.
Eligibility and recruitment
All subjects provided written informed consent.
Inclusion criteria included: (1) individuals with type 2 diabetes, (2) 18 years of age or more and (3) glycated haemoglobin (HbA1c)≥6.5%. Pregnant women were excluded from the study. Study participants were recruited with the help of district health workers who actively referred their patients.
We also recruited health workers from the district for interviews using a purposive strategy. We recruited 12 providers that: (1) had participated in trainings for the type 2 diabetes self-management education and support intervention and (2) had implemented the intervention with at least one of their patients. We interviewed two supervisors that had actively participated in the intervention. Finally, we interviewed six individuals living with diabetes. For these interviews, we organised individuals in quintiles according to change in HbA1c. Next, we selected at random three people in the first quintile and three people in the last quintile.
Quantitative data was collected at two points, originally planned for baseline and post intervention. Due to the pandemic suspending the intervention, the second data could only happen once the COVID-19 peak had passed and it was possible to contact study participants again. Data was entered in real time using data capture software (REDCap). All baseline quantitative data was collected in the home of the study participants. Due to biosecurity guidelines and in order to minimise in-person contact data collection for the second timepoint was partially collected by telephone, and only anthropometric and clinical data was collected in-person.
Qualitative data consisted of semistructured interviews with healthcare workers (n=12; 10 nurses (ages 20–45 years; 58% women); 2 supervisors (ages 40–59 years; 100% women)) and individuals living with diabetes (n=6, ages 39–70 years, 100% women). The sample size for the qualitative interviews were based on our previous experience with the implementation study in the other eight districts where we reached thematic saturation with less than 5–6 interviews.19
The interview guide was adapted from the previous study done in the eight other districts (online supplemental annex 1).19 Interviews with patients were conducted in Maya K’iche’ with the support of a research team member bilingual in Spanish and K’iche’ (MG) and an anthropologist (AA) with experience conducting interviews on health services. Interviews with healthcare workers were conducted in Spanish. Telephone interviews had a mean duration of 49 min in K’iche’ and 33 min in Spanish and were recorded and transcribed by a professional service.
We used Stata V.17.0 (StataCorp). We compared participant baseline characteristics using the proportion test for categorical data, means and SD for continuous variables with normal distributions, and medians and IQRs for continuous variables without normal distributions. In order to compare prepandemic and postpandemic results, we constructed multilevel mixed effects models for blood pressure, body mass index (BMI), HbA1c, the Diabetes Knowledge Questionnaire (DKQ-24), the Diabetes Distress Scale (DDS) and components of the Summary of Diabetes Self-Care Activities (SDSCA). As previously described, all models were adjusted for age, sex, ethnicity, education, time since diagnosis, difficulty paying for medications and initial dependent variable values.19 In order to investigate the impact of missing data, we carried out a sensitivity analysis using multiple imputation with chained equations and 100 imputations.20
We analysed interviews using Dedoose (Sociocultural Research Consultants). We conducted a thematic framework analysis using an inductive approach. First, we identified all comments that mentioned the pandemic. Second, one author (PR) developed a codebook by analysing four interviews. Third, all interviews were coded by one author (PR) while a second author (ST) reviewed all interviews and codes for accuracy and completeness. The final interviews analysed did not contribute new information indicating thematic saturation.21
We integrated our findings using a joint display of quantitative and qualitative data to facilitate conclusions and inferences from the data.22
Patient and public involvement
Patients and the public were not involved in the design, conduct, reporting or dissemination of this research.
At the start of the intervention, 157 individuals living with type 2 diabetes were referred by the district healthcare staff. Of those, 142 (90.4%) were included in the study and 15 (9.6%) were excluded because they did not meet eligibility criteria (HbA1c<6.5%). Baseline data was collected from all subjects. Endpoint data availability was: HbA1c (n=120, 85%), blood pressure (n=104, 73%), BMI (n=68, 48%), DKQ-24 (n=142, 100%), SDSCA (n=85, 60%) and DDS (n=85, 60%) (figure 1).
Table 1 details sociodemographic and clinical indicators of study participants. All participants self-identified as Indigenous Maya and the majority were women (87.3%). Mean sample age was 49.8±11.3 years. Baseline glycaemic control was poor with a mean HbA1c of 10.5±2.0 (85.2% with an HbA1c>8.0%).
Changes in clinical indicators during the pandemic
Data was collected between June 2019 and February 2021. Although endpoint HbA1c was available for 85% of subjects, the rate of missing data for other elements of the survey was much greater. Therefore, to evaluate the robustness of our primary outcomes, we conducted a sensitivity analysis using multiple imputation (table 2). The initial regression models are detailed in online supplemental table 2. After imputation the results for HbA1c were similar: 0.54 (95% CI 0.14 to 0.94), but the results for BMI were no longer significant, −0.38 (95% CI, −1.40 to 0.64).
Changes in psychometric self-care indicators during the pandemic
Study participants knowledge of diabetes measured by the DKQ-24 worsened in both the initial regression model (−3.42 (95% CI −4.46 to –2.39)), as well as the multiple imputation (−3.54 (95% CI −4.56 to –2.51)) (online supplemental table 2, table 2). Symptoms of distress caused by diabetes measured using the DDS appeared to improve, but these findings were not supported by the multiple imputation (−0.19 (95% CI −0.40 to 0.01)). There were no significant changes in self-care activities measured by the SDSCA, although the habit of following a healthy diet and exercising were low (median of 0 days per week for each activity).
Qualitative perspectives on the impact of the pandemic on the type 2 diabetes care
Thematic analysis of the semistructured interviews with district healthcare providers and individuals living with diabetes revealed nine common themes (table 3). These nine themes were organised into four areas (fear and conflict, logistic changes, supply chain and food security) and are detailed in the following text. Table 3 provides additional quotes.
Fear and community conflict
An important barrier identified by the majority of providers was communities’ general fear of contracting COVID-19 (10/13), which resulted in many study participants not visiting health services or not allowing healthcare workers to conduct home visits. For example.
Some of the patients really are afraid of home visits, “they are going to come see us and they will have COVID” and since they already realized that as individuals living with diabetes and other chronic diseases that they have more problems. With that in mind we could not continue… (Provider 7)
At the same time, more than half of the providers (7/13) emphasised that although initially during the pandemic there was a lot of fear of COVID, later the community began to reject protective measures, which generated other forms of conflict with the healthcare workers:
The people here are accustomed to not use masks so we would try to protect ourselves by wearing masks although they still put us at risk. So we were there “wear your mask, even for a little while you are with us”, so they began to look at one suspiciously, but I have to protect myself. Well that also we saw. (Provider 6)
Also, additional conflicts arose between some community members due to their opposition to the COVID-19 vaccine, which impacted other routine work flows:
… yesterday I was visiting people who were reluctant to vaccinate… we were vaccinating children and everything was going as usual but now that the COVID vaccine is out people are saying “No, I will not vaccinate my child because you want to give them the COVID vaccine.” (Provider 10)
Interestingly, these themes were not mentioned by individuals living with diabetes that we interviewed (1/6 fear, 0/6 conflict).
Logistical changes caused by the pandemic
Due to the remote location of the district, there were changes in the availability of public transportation and restrictions on the amount of time providers could spend delivering community and home-based health services. This was identified as a barrier for some providers (5/13):
The challenge that I had was getting around, getting from here to the community. The main street was closed for four months so I had to go walking to be able to do all my work in the community. It was difficult for me because I had to walk from here to the community. (Provider 8)
These changes, according to the majority of the providers (12/13), impacted their ability to carry-out home visits to deliver medications and education on diabetes. However, some providers mentioned that with planning they managed, although the visits were much briefer.
Before the pandemic the visits lasted a bit longer but now with COVID we tried to finish a bit earlier. When we covered a topic we gave some tips but didn’t expand further on it. (Provider 5)
Two individuals living with diabetes also mentioned the reduction in the number of visits that they received from healthcare workers.
…the healthcare workers always arrived to leave medications but lately they have not come… (Patient 1)
Additionally, some providers (6/13) and the majority of individuals living with diabetes (4/6) mentioned that the end of group activities (diabetes clubs) in health centres impacted the care individuals with diabetes received.
Previously in 2018 we came to work in the mornings, to make an appointment at 8 am and prepare snacks, food that follows the diet that they should have. Now, in 2020, it became difficult for us and we can’t do it. We can’t gather everyone together here. We did a few times but with the social distancing it is not the same as it was before. (Provider 2)
Another factor mentioned by a few providers (3/13) was the lack of personnel when the healthcare workers began to contract COVID-19 and have to complete quarantine. Also, a few providers (3/13) mentioned the restrictions on activities with individuals living with diabetes initially imposed by the Ministry of Health for the high-risk that COVID-19 represented for them.
…everything stopped. As I told you, remember how the quarantine was… and the diabetics and hypertensives, because of the high risk [from COVID], they were not visited due to this problem. Yes well. No activity was carried out or continued with diabetics, no activity at all. (Provider 3)
The majority of providers (10/13) mentioned that the shortage of medications for chronic diseases and other supplies (eg, glucometers and test strips) was a determining factor during the first part of the pandemic. This theme was mentioned by only one of the interviewed individuals living with diabetes.
That too, the truth is we had a little bit of problems, here, I don’t remember [exactly] how many months that we did not have medications. Until last month and this month we haven’t had [medications], we haven’t had treatment… I’m sure that at the central level there is no medicine because, our boss here, she is really great at negotiating, but still up until now we haven’t had any [medicine]. (Provider 9)
A healthy diet is an important para of diabetes control. During the first months of the pandemic, the cost of basic foods increased in Guatemala. This theme was not mentioned by any of the interviewed individuals living with diabetes but was mentioned by three providers who noted a negative impact:
I think that that part was a little impacted, for once because there almost wasn’t any work, so to put food on the family’s plate, yes that was a little impacted, that part of their diet, so [the patients] no longer followed the dietary instructions that had been given to them, whatever there is available is what they eat. (Provider 6)
Integration of data
We integrated our findings using a joint display of quantitative findings, qualitative findings and meta-inferences in order to make conclusions about the study. These are detailed in table 4.
The SARS-CoV-2 global pandemic has disproportionately impacted Indigenous populations. Although COVID-19 has had greater mortality and morbidity for individuals living with chronic diseases, the majority of studies about its impact have been in high-income countries. It is important to detail the impact of COVID-19 on chronic diseases in Latin America, a region with high levels of inequality, unstable public health systems and Indigenous populations that are experiencing epidemics of non-communicable diseases, the double burden of malnutrition and structural racism.12–14 23
In this study, we used a database of individuals with type 2 diabetes derived from a health district of the Ministry of Health in Guatemala to explore how the COVID-19 pandemic has impacted chronic disease care. Although small, this database is unique in that it pertains to an extremely poor, rural and Indigenous Maya K’iche’ community. Using regression and multiple imputation to adjust for covariables and control for missing data, we found that individuals living with type 2 diabetes showed a marked deterioration in their glycaemic control during the pandemic (change in HbA1c+0.54 (95% CI, 0.14 to 0.94), table 2) while other available clinical indicators (BMI, blood pressure) remained relatively stable. This deterioration was accompanied by a decrease in diabetes knowledge (DKQ-24, table 2).
After integrating the quantitative and qualitative findings (table 4), the most salient explanatory factor for both providers and individuals living with diabetes was decreased in-person activities by healthcare workers, which included reductions in both home visits and peer group meetings. Although medication supply and stockouts were also contributing factors, self-reported diabetes medication use remained relatively high (median of 6 days per week in both pre and post surveys, table 2, online supplemental table 2). This could be related to the existence of alternative sources of medication supply (eg, private pharmacies), although we do not have concrete data to confirm this. In addition, the close association between deteriorating diabetes indicators and limited ability to interact with front-line health workers highlights the importance of strengthening primary care systems in rural Guatemala, not only for pandemic readiness but also for overall improvements in chronic disease outcomes.24–26
These findings have important implications for public health. A recent study of the state of diabetes care in 55 low-income and middle-income countries revealed that although many people with diabetes do not receive adequate pharmacological treatment, additional deficiencies are lack of access to self-care and life-style change education.27 We have documented the marked deficiency of diabetes education in Indigenous communities in Guatemala.19 28 Also, we have demonstrated that an accompaniment and solidarity model using community health workers can improve self-care significantly, including when access to medications is inadequate.19 28 29 Conversely, when these community health workers have less contact with individuals living with diabetes, which is what happened in this health district during the pandemic, glycaemic control worsened substantially. In the face of the COVID-19 pandemic or any other health crisis, equipping and protecting front-line health workers so that they can work continuously, safely and in solidarity is a key strategy.30
This study has some strengths and limitations. First, it is a small sample derived from just one health district in Guatemala and it cannot be generalised to other countries in Latin America. However, the data fill a gap in the literature about the impact of COVID-19 on Indigenous communities in the region. Second, there is missing endpoint data in the study due to the pandemic. However, we used multiple imputation, a rigorous methodology, to address this deficiency. In addition, as is common in diabetes studies in the region, male participation was low compared with women. Finally, we were unable to track the development of COVID-19 infection or recovery in our population due to an extreme shortage of COVID-19 tests in most rural localities in Guatemala during the study period. As a result, we are unable to correlate our findings with the evolution of COVID-19 infection.
In conclusion, we presented observational results on the impact of COVID-19 on type 2 diabetes care in an Indigenous community in Guatemala. The main finding is that glycaemic control and diabetes knowledge worsened, likely attributed to the loss of regular contact between the health system and opportunities for health education in peer groups and with healthcare workers.
Patient consent for publication
This study involves human participants and was approved by the Maya Health Alliance (WK 2018 005) and the Institution of Central America and Panama (CIE-REV 082/2018). Participants gave informed consent to participate in the study before taking part.
This post was originally published on https://bmjopen.bmj.com