Introduction

Diabetes mellitus is a chronic, non-communicable disease (NCD) with an increasing public health burden disproportionately affecting those in low resource settings. In Africa, incidence of diabetes is expected to increase by 129% to 55 million in 2045 [1]. Sustained glycemic control significantly reduces risk for early morbidity and mortality from diabetes [2], yet the majority of people living with diabetes (PLWD) do not meet globally defined targets. People living in low and lower-middle income countries (LLMICs), especially rural areas, face significant barriers to achieving glycemic goals. In rural Liberia, specific barriers include: the scarcity and high cost of medications and appropriate food, the need to travel far distances to receive care without transportation reimbursement, the lack of awareness about diabetes, and the effect of diabetes stigma in the community [3]. These challenges are often exacerbated in people living with insulin dependent diabetes (PLWIDD). Thus, it is even more essential that patients have a strong clinical understanding and problem-solving skills in this setting.

To improve care for PLWIDD living in rural areas, in 2017 Liberia initiated the PEN-Plus strategy of care for severe chronic conditions beginning with Maryland County in the impoverished Southeastern part of the country. PEN-Plus (the Package of Essential Noncommunicable Disease Interventions – Plus) is a comprehensive strategy to decentralize care for multiple severe NCDs including insulin-dependent diabetes by training and equipping mid-level providers to manage several chronic conditions using shared infrastructure [4,5]. In 2022, all 47 countries of the World Health Organization’s African region adopted a resolution to initiate PEN-Plus programs by 2030 [6,7].

The goal of PEN-Plus programs in this setting is to achieve an “intermediate” level of care to reduce the risk of morbidity and mortality from diabetes. “Intermediate” care involves two to four blood glucose checks per day, multiple daily injections of insulin, availability of Hemoglobin A1C (HbA1c) testing and other complication screening, and access to diabetes education with social support [8]. PEN-Plus programs, including the program in Liberia, have access to insulin, blood glucose strips, and HbA1c testing, but may not have formalized plans for providing diabetes education, addressing the psychosocial burden of living with diabetes, or empowering and engaging patients with their own care.

Diabetes self-management education (DSME) is an evidence-based, structured method to educate and empower PLWD to improve self-care by focusing on the behavioral aspects of diabetes management and problem-solving to address barriers to self-management [9]. The Association of Diabetes Care and Education Specialists (ADCES) defines seven key self-management behaviors: healthy eating, being active, monitoring, taking medication, problem-solving, healthy coping, and reducing risks [10,11]. The theoretical basis for DSME draws from several behavior change models, including the health belief model [12], social cognitive theory [13], and the transtheoretical model [14]. Practically, DSME involves a shift in clinic culture to center patient education and psychosocial needs. Specific tools used in DSME include use of non-judgmental language (i.e., “person with diabetes” instead of “diabetic”) [15], motivational interviewing, and SMART (Specific, Measurable, Achievable, Relevant, Time-bound) goal setting.

In high resource settings, DSME has been shown to improve diabetes knowledge, self-management behaviors, clinical outcomes, and psychosocial wellbeing while also reducing cost to healthcare systems [16,17]. However, research in LLMICs remains limited and focused on type 2 diabetes (T2D). A recent scoping review based in Africa evaluated 19 studies about DSME inclusive of patients with type 1 diabetes (T1D), T2D, or both. Of the four studies that evaluated self-management behaviors, none found positive impact. However, 10 of the 14 studies evaluating change in HbA1c found positive impact [18]. These findings are supported by a systematic review by Lamptey et al, which found an improvement in HbA1c in all nine studies evaluated ranging from 0.5% to 2.6% [19]. Several randomized control trials found similar reduction in HbA1c, but did not measure behavioral or psychosocial outcomes [20,21]. To date, no studies of DSME have looked specifically at PLWIDD, feasibility outcomes, or psychosocial impact of DSME in LLMICs.

This study aims to:

1. Evaluate change in provider knowledge of DSME delivery through training and mentorship.
   1. Assess the feasibility of providers delivering DSME to PLWIDD in a rural region in Liberia.
   2. Measure the effect of DSME on self-management behaviors, psychosocial well-being and clinical outcomes among PLWIDD.
   3. Determine acceptability of DSME among providers and PLWIDD.

Materials and methods

Data collection and analysis

Qualitative data

Semi-structured qualitative interviews were conducted by trained research staff members. Boston-based researchers completed interviews with providers in English. Liberia-based study staff members trained in qualitative interview techniques interviewed PLWIDD in Liberian English. Interview topics with providers and patients are described in detail in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Interview topics.

https://doi.org/10.1371/journal.pgph.0005657.t001

Interviews were audio recorded and transcribed by two study staff. Interviews with PLWIDD were transcribed by a study member familiar with Liberian English and directly translated into English. A codebook was developed based on interview guides, and transcripts were independently double coded using thematic analysis. Qualitative coding and analysis were done using Dedoose [24]. Analysis utilized two iterative steps. First, we utilized an a-priori thematic analysis. Then, after the first round of coding we added additional themes that had emerged. Emerging themes were discussed between coders to increase validation. One of the two researchers who participated in the coding and analysis is based in Boston and has no affiliation with the communities or health facilities where study participants were recruited. The second researcher was involved in training providers on DSME.

Quantitative analysis

All quantitative data analysis was completed in STATA v 15.1 and R version 4.2.2.

HbA1c.

Due to procurement issues with HbA1c cartridges, no HbA1cs were checked at month nine. Some individuals were missing measurements for particular time points; estimates for all available data and for the subset of individuals without missing measurements are shown at any time point. To account for repeated measures, estimated changes in HbA1c calculated from baseline using a mixed effects model are shown below. HbA1c was estimated for individual i at time t with fixed effects (β_1–3) on time indicators (T3, T6, and T12) for the three-, six-, and 12-month measurements, a fixed intercept for baseline (β₀), and a random intercept for individuals (δ_i).

Knowledge assessments.

A 41-question multiple choice exam was developed to assess DSME skills and knowledge of diabetes management. This assessment was designed to match pre-existing competencies for diabetes management within PEN-Plus, so to date has not been externally validated. Domains included psychosocial management, age-appropriate education, basic diabetes pathophysiology and management, glucose levels, glucose monitoring, insulin administration, insulin dosing, nutrition and diet, physical activity, and complications screening.

Patient knowledge assessments were developed in alignment with the ADCES seven self-care behaviors [9] and adapted with guidance from local providers to local context in Liberia. Because most patients have limited literacy, providers read the questions to the patients and marked their answers.

The mean and standard deviation of the scores on assessments were calculated at each time point. To account for repeated measures, score changes were estimated from baseline using a mixed effects model shown below, where Y is the knowledge score for individual i at time t, T6 and T12 are indicators for the six- month and 12-month assessments, respectively, and δ_i is a random intercept for individuals. Coefficients β_1–2 are fixed coefficients, and β₀ is a fixed intercept.

Results

Demographics

Five providers were enrolled who participated in the initial training and were interviewed at baseline. Of these five, the three providers who regularly staff the NCD clinics at JJD and PHC were followed for the entire duration of the study. The other two providers who were enrolled at the beginning were not followed for the entirety of the study due to conflicting clinical responsibilities outside of NCD clinics

There were 26 PLWIDD enrolled in care at the two clinics at the onset of the study. All 26 PLWIDD were eligible, agreed to participate and provided informed consent. Eleven were from JJD while 15 were from PHC. Fourteen were female and 12 were male. The average age of participants was 34 years (range 17, 67), and eight (33%) were married. Level of education and employment varied across participants (Table 2). Twenty-four participants were living with T1D, while two participants had insulin-dependent T2D. Twelve participants were using NPH and Regular insulin, while 14 were using NPH only. The average diabetes duration was 4.3 years. The majority (81%) were receiving food or financial assistance (Table 2). One participant was withdrawn from the study during month three due to experiencing severe mental health challenges resulting in inability to complete study activities. One participant died in month 10.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Participant sociodemographic characteristics.

https://doi.org/10.1371/journal.pgph.0005657.t002

Primary outcomes

Acceptability.

Overall, patients and providers found DSME to be acceptable and superior to how diabetes was previously managed. Providers reflected on how receiving DSME training allowed them to gain a deeper understanding of patient’s lived experiences with diabetes, changed their perspective on the day-to-day difficulties of diabetes management and gave them a greater sense of compassion for PLWIDD. One provider discussed specifically how their own practice has changed since receiving DSME training (Table 3, quote 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Qualitative interview quotes.

https://doi.org/10.1371/journal.pgph.0005657.t003

Despite it being perceived as a positive shift, it took time for PLWIDD to feel comfortable with these changes (Table 3, quote 2). Providers reported that patients felt they now had autonomy to make decisions about their care, which is unfamiliar in the traditional health system (Table 3, quote 2). PLWIDD also felt a main benefit of DSME was the shift in interactions with providers, substantiating some of the self-reflections described by providers (Table 3, quote 3). Overall study findings show that DSME had the unexpected benefit of improving the relationship between PLWIDD and their providers.

Many PLWIDD conveyed how DSME had led to an increase in knowledge and self-confidence in their ability to manage their condition (Table 3, quote 4). PLWIDD reported feeling empowered to actively participate in improving their health outcomes. (Table 3, quote 3).

Many participants mentioned more frequent engagement in self-management behaviors, especially SMBG and insulin injections. One participant noted how nutrition education, particularly the timing of meals with insulin doses, had helped with blood sugar levels. Some were proud of the way they were managing diabetes since the implementation of DSME, and others noticed their blood sugars were within range more often. Additionally, the opportunity to interact with other PLWIDD increased trust in the medical team. They particularly appreciated group education sessions as they facilitated peer learning (Table 3, quote 5).

Adoption.

The implementation of DSME had challenges, particularly in training and implementation. Training related challenges included logistical challenges due to COVID-19, shortage of staff and growing patient load, scheduling and time constraints, and introduction to SMART goal setting (Table 3, quote 6). SMART goal setting was challenging to initiate with patients due to its novelty, but with more exposure over time patients started to gain a better understanding. Providers also reported that as patients become more comfortable with SMART goal setting, the benefits of DSME became noticeable (Table 3, quote 7).

Despite the challenges, providers felt DSME was sustainable (Table 3, quote 8) and provided recommendations on how to enhance the feasibility and adoption of DSME by incorporating other education strategies, including visuals. Their active involvement reflects a keen interest in the project and their commitment to integrating DSME into their future practice.

Fidelity.

A total of 272 DSME monthly clinical implementation forms were analyzed. There were no major fluctuations by month in the number of monthly DSME forms completed. However, the frequency at which providers reviewed SMART goals from the previous visit varied from 39.1% to 100% (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Review of SMART Goal Frequency.

https://doi.org/10.1371/journal.pgph.0005657.g001

Mentoring checklists completed by clinical mentors during patient simulations were used to assess utilization of DSME to guide patient encounters. All providers asked about patients’ biggest concern for that visit. SMART goals set at the previous visit were reviewed in five out of six simulated patient encounters. Teach-back was evaluated only during the first simulation, and only one of the three providers used this method. Providers involved patients in decision-making regarding their treatment plan for all six simulated patient encounters observed.

Frequency of self-management behaviors.

Participants brought glucometers to clinic at 87.1% of encounters and brought logbooks to clinic 59.6% of encounters. Average blood glucose checks increased from 4.0 (95% CI 1.93, 6.23) checks per week month one, to 7.2 (95% CI 3.56, 10.79) checks per week month 12 (Fig 2). Average missed insulin injections decreased from 3.9 (95% CI 2.51, 6.03) in month one, to 0.82 (95% CI -0.14, 1.79) in month 12 (Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Patient self-management behaviors (blood glucose checks and missed insulin injections) recorded in the week preceding clinic visits by month.

https://doi.org/10.1371/journal.pgph.0005657.g002

Secondary outcomes

HbA1c.

Mean HbA1c at baseline was 12.0% (107.7 mmol/mol) (n = 26, se 0.41). At three and six months, HbA1c was 9.1% (76.0 mmol/mol) (n = 19, se 0.64) and 10.9% (95.6 mmol/mol) (n = 17, se 0.40) respectively. At 12 months, HbA1c was 10.9% (95.6 mmol/mol) (n = 24, se 0.49) with a modeled mean reduction from baseline of 1.1% (95% CI 0.0 to 2.0) (Fig 3).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Mean A1C by month, for all participants and participants who completed all study visits.

https://doi.org/10.1371/journal.pgph.0005657.g003

Severe adverse events.

Over the course of the year there were five severe adverse events. There were three episodes of DKA that occurred during months 3, 6, and 8. One patient experienced hypoglycemia requiring assistance during month 7, and there was one participant death in month 10 from excessive alcohol intake.

Provider knowledge.

Providers performed better on the knowledge assessment at six months compared to baseline. The average score increased from 60.2% (n = 3) prior to DSME training to 71.5% (n = 3) at month six. Due to travel restrictions, provider knowledge was not assessed during month 12.

Patient knowledge.

Scores on the patient knowledge assessment increased from a mean of 32.4% (n = 26, se 4.3) at baseline to 58.1% (n = 25, se 5.7) at month six, a modeled increase from baseline of 25.0% (95% CI 15.0 to 35.1%) accounting for the slightly different composition of individuals with baseline and six-month follow-up data. At month 12, scores increased to 64.6% (n = 21, se 5.2), with an estimated mean increase from baseline of 33.7% (95% CI 23.0 to 44.3%). (Table 4)

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Patient knowledge assessment scores over time.

https://doi.org/10.1371/journal.pgph.0005657.t004

Discussion

This pilot study found DSME to be acceptable and feasible while having a positive impact on clinical outcomes. The intervention was supported by input from the population of PLWIDD through the establishment of a PAB. Regular PAB meetings informed implementation decisions and encouraged patient involvement and engagement in the study. Qualitative analysis suggests DSME was well accepted and useful for providers and PLWIDD alike. DSME was associated with improved patient knowledge, frequency of self-management behaviors, and HbA1c in PLWIDD. Patient knowledge assessment scores had a mean increase from baseline of 33.7% (95% CI 23.0 to 44.3%), suggesting retention of new knowledge learned throughout the DSME process. This knowledge increase was associated with an increase in SMBG and decrease in missed insulin injections. Other DSME programs in sub-Saharan Africa (SSA) examining behavioral outcomes have not shown an impact on SMBG or medication adherence, however, these studies primarily enrolled people living with type 2 diabetes and were conducted in urban areas or tertiary hospital settings [18,25,26].

Participants brought glucometers and logbooks to clinic most visits, 87.1% and 59.6% of all encounters respectively. This was less than a study conducted in a similar setting in rural Malawi, which found participants brought glucometers and logbooks to clinic 98% of visits [27]. However, the study in Malawi was specifically evaluating SMBG implementation, and more effort may have been spent reminding patients to bring supplies to clinic visits. The increases in knowledge and self-management behaviors were associated with a 1.1% decrease in HbA1c from baseline to endline. Of note, the largest observed decrease was in month three, followed by an increase in month six. This may have been the result of seasonal variability in food security, which can negatively impact diabetes management or from fatigue in adherence to protocols from providers or patients over time. Due to the small sample size, more robust studies will be needed to confirm these results. These findings do align with other studies evaluating DSME interventions in SSA, which have reported decreases in HbA1c ranging from 0.5% to 2.6% [19].

An additional noteworthy finding was the perceived cultural shift to patient-centered care. Traditionally, in Liberia, providers are seen as authority figures whose plan of care needs to be followed, without eliciting the engagement of patients. Even though it took time for PLWIDD to understand the concept of setting goals and to feel comfortable participating in the development of their care plan, results suggest that shared decision-making was feasible even in settings where patients have limited education, health literacy, and resources. Not only was this shift in patient-provider interaction notable by both providers and PLWIDD, but so was the increase in self-confidence in patients’ knowledge of diabetes and ability to manage it, increased problem-solving skills around diabetes management, and an increase in well-being and satisfaction with their care.

This study has several limitations. While the entire known population of eligible PLWIDD in Maryland county were enrolled, this totaled only 26 participants. Because there was no control group, it was not possible to show a causal relationship between the intervention and changes in outcomes. Patient knowledge assessment and psychosocial assessment were administered by providers, which may have led to participants not answering honestly. Qualitative results from the study should be interpreted cautiously due to limited literacy from some PLWIDD and risk of bias as interviews were conducted by providers who provide regular clinical care. Change in self-management behaviors were evaluated by looking at the week preceding each clinic visit, which may have over-estimated self-management behaviors. Finally, the site where the study was conducted is supported by the organization Partners In Health, which can provide medications and diabetes supplies when there are local supply chain issues. This may limit the generalizability of results to other rural clinics, where supply chain challenges can be a major barrier to diabetes care [3,28].

This study highlights several key areas for future research. Repetition of this study with a more robust design and larger sample size would be beneficial. In general, studies evaluating DSME or other behavioral interventions should consider including knowledge, behavioral, and psychosocial outcomes in addition to clinical outcomes. Though we were not able to report on psychosocial outcomes from this study due to data quality issues, we encourage inclusion of these outcomes in future research. While this study focused on diabetes, it was conducted in a clinic utilizing the PEN-Plus model, where providers also care for people living with other serious NCDs [4,29]. Other NCDs often require education and support such as healthy lifestyle counseling. Further, many of the competencies learned by providers through DSME are applicable to providing education and support to patients living with other NCDs. Further research expanding the scope of DSME to other conditions, especially in low resource settings implementing integrated care models such as PEN-Plus is warranted. Inclusion of PLWIDD themselves, whether through a PAB or peer support groups, is essential.

Conclusion

This pilot study showed that a structured DSME program in a low resource, SSA setting was feasible, acceptable, and associated with increased patient knowledge, an improvement in self-management behaviors, and improvement in HbA1c. Further research, including expanding the scope of DSME to include other NCDs is warranted in the context of PEN-Plus expansion. As the first to evaluate DSME among PLWIDD in a rural, LLMIC setting, this study adds to the growing body of literature showing that the combination of support and education provided by DSME can address many of the barriers to care faced by PLWD globally.

Supporting information

Acknowledgments

Thank you to staff and colleagues that supported the study in various capacities. A special thanks to the PAB for advising us through this process and keeping patients’ perspectives at the forefront, and to the Liberia MOH and PIH Liberia for allowing us to conduct this project at their facilities. We are appreciative of Carolina Noya, FNP, PhD and Maureen McGrath, RN, MS, PNP for their ongoing mentorship around teaching, training, and study design. In memory of Cyrus Randolph and his tireless commitment to helping others as a provider, mentor, colleague, and friend.