Implementation and evaluation of a team approach to managing diabetes mellitus and diabetic retinopathy in the South African district health system

was necessary in 36.5 % of cases. Conclusion: The implementation of a team approach to co-manage DM and DR utilising the developed protocol improved the quality of care for patients with DM in the district health system.


Introduction
Diabetes mellitus (DM) is a global pandemic 1 affecting patients in their most productive years. It is a lifestyle disease, unless genetically inherited, associated with modifiable risk factors such as hypertension and obesity, and non-modifiable risk factors such as ageing, ethnicity and family history. 2 The microvascular and macrovascular complications of DM impact on patients' quality of life, contributing to high morbidity and mortality rates.
Microvascular and macrovascular complications in DM may develop simultaneously. 3 Complications include cerebrovascular disease, cardiovascular disease (CVD), peripheral vascular disease, nephropathy, neuropathy and diabetic retinopathy (DR). The microvascular complications of DM are governed by the polyol pathway, which is activated by persistent hyperglycaemia. 4 Hyperglycaemia is measured using an haemoglobin A1c (HbA1c) test, which assesses the average glucose level of the previous 2-3 month period. 5 Poor HbA1c levels have been associated with the development and progression of DM complications. 6 Macrovascular complications of DM have been found to be associated with insulin resistance more than hyperglycaemia. 7 High insulin resistance has been found to increase CVD risk by 2.5-fold. 8 from simple steatosis to non-alcoholic steatohepatitis to cirrhosis and hepatocellular carcinoma. 9 Type 2 DM and obesity have been identified as risk factors for the development of NAFLD. 10 It has been established that CVD risk increases by 1.87-fold in patients with NAFLD in the presence of Type 2 DM. 11 Non-alcoholic fatty liver disease also increases microvascular complications of DM such as chronic kidney disease and DR. 12 The gold standard for diagnosing NAFLD is a liver biopsy to identify liver steatosis and carcinoma. However, in the absence of liver steatosis, the recommended treatment is a weight loss and exercise programme to enhance insulin sensitivity to promote the reduction of liver steatosis. 13 Diabetic retinopathy, cataract and glaucoma are common ocular manifestations of DM. 14 Uncorrected refractive errors (myopia, hyperopia and astigmatism) are generally a major cause of low vision and blindness. Low vision is defined by two categories, namely moderate visual impairment (VI), which is presented by visual acuity (VA) less than 6/18 but equal to or better than 6/60 and severe VI, which is presented by VA less than 6/60 but equal to or better than 3/60. 15 Factors responsible for uncorrected refractive errors include the lack of awareness and recognition of the problem, nonavailability of refractive services or inability to afford refractive services and cultural disincentives to compliance. 15 Refractive changes in patients with DM can be acute or chronic or progressive as associated with an increase in myopia. 16 An increase in crystalline lens thickness is known to increase the prevalence of late-onset myopia in patients with DM. 17,18 Diabetic retinopathy has the potential to affect all patients with DM and severe DR has been associated with the presence of neuropathy and nephropathy. 19 Patients with DR may not experience symptoms of vision loss until it becomes too late to treat, leading to irreversible blindness. Some patients with DR may only notice vision deterioration if they develop cataracts. Cataract is the most common ocular complication of DM and is considered a major cause of VI in patients with DM in developed and developing countries. 20 Patients with DM are 4-5 times more likely to develop cataracts with an early onset. 21 Despite advancement techniques for cataract surgery (phacoemulsification), poor VA following cataract surgery has been associated with posterior capsular opacification, post-operative cystoid macular oedema, diabetic macular oedema (DME) and worsening of DR in patients with DM. 20 Patients with DM have an increased risk of developing primary open angle glaucoma (POAG). 22 A glaucoma suspect is defined as an individual with clinical findings and/or a constellation of risk factors that indicate an increased likelihood of developing POAG. 23 These clinical findings and risk factors range from having high intraocular pressure (IOP), optic nerve head or retinal nerve fibre layer (RNFL) appearances suggestive of glaucomatous damage, unexplained visual field defects and abnormal angles to a positive family history of glaucoma. 24,25 An IOP greater than 21 mmHg was traditionally set as a criterion for the suspicion of glaucoma. 26 Other studies have also used this criterion. 14,25 It was observed from the study of Baisakhiya et al. 27 that patients with poor glycaemic control had higher IOPs and Hymowitz et al. revealed that it was unlikely for patients with elevated HbA1c levels to have low IOPs. 28 Ideally, in the holistic co-management of patients with DM, HCPs should be educating patients on systemic and ocular complications, which may arise when the DM is not adequately controlled. With the rising incidence of DM, there is a growing focus by HCPs towards assisting patients with DM management because self-management poses a challenge. 29 Values for DM detection have been documented, yet these values may appear meaningless to the illiterate or uneducated and may be underestimated by them.
As a result of the asymptomatic nature of DR, early detection of DR is essential to prevent further progression and potential vision loss. Health education is meant to be promoted at each level of care in the district health systems (DHS) involving primary healthcare (PHC) nurses, medical officers (MOs), ophthalmic nurses and optometrists. However, concerns have been expressed regarding the lack of policies dedicated solely to eye health promotion in South Africa (SA) where policies have been dedicated to curative rather than preventative measures. 30 Patients with DM require comprehensive management by a team of HCPs to manage both systemic and ocular complications to prevent and reduce the progression of blindness because of diabetic eye disease. This highlights the need for a DR screening model in the DHS to co-manage patients with DM and DR.

Methodology
An observational study using convenience sampling was employed whereby 201 patients with DM from the seven catchment area clinics for Voortrekker district hospital were recruited for participation from 01 October 2018 to 31 March 2019. From 201 recruited patients, 107 consented to participation ( Table 1). The study made use of the input, processes and output approach (Appendix 1). The duration of the retinal screening process was from 08 April 2019 to 30 May 2019.
Visual acuity was assessed using a Snellen VA chart. A Tumbling E chart was used to assess VA of illiterate patients. A handheld autorefractor (measuring distance of 1 m ± 5 cm), known as the Eurofractor, was used to determine refractive error of patients with VA less than 6/9. The refractive error was subjectively refined using trial frame refraction. The i-Care tonometer was used to assess IOP and the Tomey nonmydriatic fundus camera (model TFC-1000 with resolution of 4096 × 3072 mega pixels and a field of view of 45 up to 80°) was used to capture retinal images.
After consenting to participation, patients were given appointment cards for their retinal screening dates by the PHC nurses and referred to the MO for blood testing and medical assessment. Transport arrangements were made for patients to be transported from the clinics to Voortrekker district hospital for the ocular screening assessment. Patients profile details (gender and age), case history (duration of DM, presence of hypertension, medication compliancy, presence of a meal plan, presence of neuropathy and family history of DM) and vital signs (weight, glycaemia, blood pressure and HbA1c levels) were documented on the record card by the PHC nurses prior to patients' arrival for retinal screening. Blood tests were also performed and results were analysed by the MO who recorded findings of cardiovascular, kidney and liver dysfunction on the record card prior to the retinal screening process. On patients' arrival, the optometrist (principal investigator) conducted the ocular screening tests as per protocol (VA assessment, pinhole test, refractive error, IOP assessment, retinal photography, DR screening and grading). Retinal photography images were graded using a combination of the Scottish Grading Classification System and United Kingdom National Grading Classification, whereas other ocular lesions were documented on the record card under other lesions (OL). Photographic images of both eyes of each patient were saved to a flash drive which was hand-delivered to the ophthalmologist at Mankweng hospital for DR grade verification and the verification of OL (such as cataract and glaucoma suspects). All retinal photographs were assessed and patients requiring referrals to the ophthalmologists were given appointment dates for referral to either Mankweng or FH Odendaal hospitals based on the equipment and treatment available at those institutions.
On completion of the screening process, satisfaction survey forms were distributed to all the HCPs involved in the screening programme to evaluate the screening process. These included 107 patients, 14 PHC nurses, one MO and one ophthalmologist. A modification of the approach was performed after conducting a strength, weakness, opportunity and threat (SWOT) analysis of the screening process. Based on the responses, recommendations were made to support the DR screening model. Data were analysed using standard descriptive statistics (proportions and frequencies) to describe all categorical variables and mean with standard deviations for continuous variables. The Fisher's test was used to test for associations between categorical variables. Spearman's rank correlation coefficient tests were used to measure correlations between continuous variables and the Mann-Whitney U tests were used to compare continuous variables between two groups because of non-normality and small samples.

Ethical consideration
Approval to conduct the study was received from the Biomedical Research Ethics Committee, University of KwaZulu-Natal (BREC reference no. 272/18). Ethical clearance was granted before commencement of the study.
Permission to conduct the research study was obtained from the Department of Health Limpopo Province and the respective Heads of Institutions at clinics and hospitals. Information relating to the study was provided and informed consent was obtained from all participants.

Results of implementation phase Discussion
Poor glycaemic control has been identified as an important predictor to the development and progression of DR. 31 In this study, a correlation between the hyperglycaemic level and HbA1c level was found (Spearman's rho = 0.58). Although 98.1% of patients were supposedly compliant with treatment for DM as seen in Figure 1, 77.3% reflected HbA1c levels of > 7% showing poor control of DM. Only 8.4% patients complied with proper meal planning. There was an association found between HbA1c levels greater than 7% and hyperglycaemia levels greater than 10 mmol/L (Fisher's exact = 0.001; Table  2). Cardiac markers to detect cardiovascular disorders and liver function test assessments were done on 64 out of 107 (59.8%) patients. Of the 64 patients, 57 (89.1%) suffered liver disorders, and this was unlikely due to NAFLD because of the low prevalence of obesity (5.6%). It was, however, not established whether the high prevalence of liver disorders was because of alcohol abuse and/or nutritional deficiencies. Creatinine levels to assess kidney dysfunction were done on 55 out of 107 (51.4%) patients.
Refraction was conducted in patients with VAs less than 6/9 (32.7%) and myopia was the most common refractive error found. The study conducted by Gwinup and Villarreal 18 established that patients with DM developed more myopia after the injection of glucose, which was because of the accumulation of by-products of glucose metabolism within the crystalline lens causing lens swelling and inducing myopia. An acute decrease in plasma glucose concentration has, therefore, been associated with hyperopia. 18 In this study, of the 14 patients who had severe  visual impairment (SVI), 71.4% had myopia in the right eye and 75% in the left eye.
A total of 16 (14.9%) patients were identified as glaucoma suspects. Of the 16 patients, eight (50%) had IOP > 21 mmHg (Table 3), whereas the other eight displayed optic discs or RNFL appearances suspicious of glaucomatous damage. In these patients, the ISNT rule (measurements of disc rim thickness of inferior ≥ superior ≥ nasal ≥ temporal) 32 was not obeyed. No correlation between IOP and HbA1C was found (Spearman's rho = 0.27 for right eye and 0.18 for left eye). However, of the eight patients who had IOPs greater than 21 mmHg, six (75%) had HbA1c levels of more than 7% and this finding concurred with the studies of Baisakhiya et al. 27 and Hymowitz et al. 28 showing that patients with poor glycaemic control have higher IOPs. The HbA1c level in the other two patients was not assessed.
There was no association found between hypertension and glaucoma suspects (Fishers exact test; p = 1.000 for right eye and p = 0.756 for the left eye) as seen in Table 4. The ocular complications co-existing with the glaucoma suspects included cataracts in three (18.8%) patients, DR in two (12.5%) patients and hypertensive retinopathy in one (6.2%) patient.
Cataract was the most common ocular complication found (Table 5) causing blindness in six out of 11 (54.6%) patients in the right eye and five out of 14 (35.7%) in the left eye. Longer duration of DM has been associated with an increased frequency of cortical cataract and increased frequency of cataract surgery. 33 In this study, there was no association found between the duration of DM and presence of cataract (Mann-Whitney U test; p = 0.65 in the right eye and p = 0.32 in the left eye). The median (IQR) duration of DM in the group with cataracts was 10 (7-12) and 10 (6-10) in the group without cataracts for the right eyes. For the left eyes, the medians (IQR) of 10 (6-10) and 8 (5-10) were found in the groups with and without cataracts, respectively.
Two of the five patients with DR (40%) and three of the four (75%) with DME in the right eye showed poor control of their glycaemic level (HbA1c > 7%). The HbA1c level of the other three patients with DR and one with DME was not tested. In the left eye, three of the five patients with DR (60%) and two of the five with DME (40%) showed poor glycaemic control. The HbA1c level of the two patients with DR and two with DME was not tested and only one patient with DME had an HbA1c level less than 7%.    Patients with pre-existing DME were found to have the worst prognosis for visual rehabilitation after cataract surgery. 34,35 Treatment guidelines have been recommended for pre-existing DR and DME, which indicate annual screening for no DR and for mild non-proliferative DR (NPDR). 36 A study conducted at a district hospital in SA revealed that 42.9% of patients with DM had refractive disorders and also confirmed that suspected glaucoma, cataracts and DR were prevalent eye conditions amongst patients with DM. 37 From the patients' satisfaction survey in Table 6, the two patients who experienced transport problems utilised their own transport to the hospital for retinal screening. For one patient, this was by choice but the other missed the transport from the clinic because of late arrival and incurred a R50.00 travel cost. One patient waited for a long time at the clinic to be transported to the hospital for eye screening as she presented before 06:00 at the clinic and transport was scheduled for 09:00. More patients requested additional education on wound healing, foot care, dental care and blood pressure control than on diet and hyperglycaemia control. Some patients reported removing their own teeth and assumed that the use of false teeth does not require dental care. Additional information requested was on weight loss, urinary incontinence, body rash, management of headaches, hearing loss and advice on social grants because of unemployment. They also complained about suffering from other ailments such as cough, shortness of breath, swollen legs and getting chest irritation after taking ibuprofen. Other concerns raised were the unavailability of insulin injections, hypertension and DM treatment on certain days at clinics. In addition, they complained that on some days, blood pressure machines and glucometers were not working and occasionally there were no glucose test strips. Another challenge was to get pills on time because of stock unavailability and misunderstanding of the dosages of alternative treatments prescribed when regular prescribed treatment was out of stock. At times no eye treatment was available at the clinics and hospital. There was not enough seating place at the clinics. A few reported that it was a challenge when it rained as they sit outside and when there is no shade this brings on headaches.
From the HCPs satisfaction survey in Table 7, the PHC nurses reported that the screening programme enhanced their knowledge and the knowledge of patients with DM. It was, however, a challenge to recall patients as their cell numbers were either on voicemail or no longer operational. At times, different test tubes were not available to collect blood samples. Patients were also not always compliant with treatment because of the unavailability of insulin injections at times. Occasionally, pills for high blood and DM were also not available. They suggested that the DR screening protocol be implemented for better service delivery as patients appreciated the opportunity of seeing the 'pictures of their eyes' for the first time. A staff member who was also a participant reported that the research was very helpful because of her own diabetic history. It was the first time that her blood was taken and tested. She was always worried about other undiagnosed conditions and never knew what her creatinine level was but now through this research, she had this opportunity to find out. Her sister passed on because of late diagnosis of liver disease; she was on dialysis after suffering from kidney failure. Another PHC nurse also reported that relatives of a patient were very pleased about the DR screening protocol as the patients' eye was operated within 2 months after being screened. The patient could now see better and manage to do things independently. The MO appreciated the effort demonstrated by the principal investigator to implement a team approach for early detection of DR.  The ophthalmologist reported seeing more than 100 DR cases per month and doing 60-70 laser photocoagulation treatments per month. The ophthalmologist was satisfied with the photographic images of the fundi and revealed that 80% of images were appropriately graded by the optometrist. She reported having sufficient space to examine and manage DR cases and also having treatment available for DR and DME. However, she reported not having sufficient staff to assist with DR cases and her request was for more ophthalmologists to be appointed to assist with DR case management.

Strength, weakness, opportunity and threat analysis
Strength: Staff revealed that patients were satisfied with the screening programme and wanted this programme to continue. There was an influx of patients making enquiries at the clinic about the booking system for retinal photography after the screening programme was completed. Patients reported never having been screened with a camera and appreciated the time taken on the additional education given on diet, foot care and dental care. The project's success was supported by the PHC nurses, MO on outreach, ophthalmologist and laboratory assistants. The PHC nurses saw the relevance of having a formal database for patients with DM. The MO in collaboration with the PHC nurses conducted the blood testing, which was sent to the laboratory at Voortrekker hospital for analysis and the progress could be easily tracked as the laboratory was on the premises of Voortrekker hospital.

Weakness:
The commitment to provide a diabetic patient database from the data capturers working at the clinics was poor. In the absence of a diabetic patient database, the PHC nurses only recruited patients presenting at the clinic over a period of 6 months (October 2018 to Mach 2019). Patients' contact numbers on their files were non-existent and unreliable making it difficult to recruit more patients. A few blood pressure readings were not taken at the clinics because of nonoperational batteries. Blood samples were not taken of every patient because of the unavailability of test tubes on some screening dates. Some patients did not comply with going back to the clinics for retesting of blood. Certain markers that could not be analysed because of the restrictions in the laboratory facility had to be sent to the laboratory in Polokwane for analysis and a few blood samples were misplaced. Body mass index (BMI) could not be calculated by the PHC nurses because only weight was measured and not height.

Opportunity:
Retinal photography created an opportunity to motivate patients for better glycaemic control. The goals of the screening programme are in line with those of the World Health Organisation (WHO), International Diabetes Federation (IDF) and National and Provincial Departments of Health. The programmes innovation could be replicated in other provinces and other parts of Africa. The programme offers an opportunity for training and employment for optometrists and ophthalmic nurses.
Threat: As a result of the lack of staff in the department, the screening programme was conducted by one optometrist who was the principal investigator.

Conclusion
The project showed feasibility to screen for DM and DR in a district health setting of SA. The screening model demonstrated improvement in quality of care of patients with DM and confirmed the importance and relevance of screening patients with DM to prevent development and progression of DR.

Recommendations
There should be more commitment from data capturers to organise databases for all patients with chronic illnesses. They should also capture more than one contact number of patients so that family members can be contacted if patients do not follow their review appointments. This will ensure better control and efficiency in follow-up to manage patients with DM. The involvement of community health workers will be beneficial to assist with patient recruitment and education on the importance of being tested for DM. This can be done by arranging campaigns highlighting symptoms and the associated complications of DM. The involvement of dieticians to assist and train PHC nurses to calculate BMI and educate patients on the importance of following proper meal plans will also be beneficial. Measuring scales and meter rulers for height assessment must be available at the clinics to determine BMI for obesity detection. The importance of blood testing should be emphasised to patients so that they see the need to be assessed for complications associated with DM. Furthermore, PHC nurses are to ensure that testing supplies are available and in good working condition for patient care at clinics, that is, blood pressure machines, glucometers, spare batteries and various blood testing tubes. Remodelling of clinics to create more working space and larger patient waiting areas will ensure more efficiency in the workplace. Staff recruitment needs to happen so that the DR screening model can be implemented by a team of HCPs dedicated to the programme.
The main aim in primary care is not to reach a specific diagnosis but to detect patients who are eligible for referral to be further managed. Healthcare practitioners (HCP) have to develop confidence in managing patients with DM and this can be achieved by engaging in group discussions with other HCPs, and by attending training courses for updates in management strategies and cascading information to other team members. Training programmes in retinal photography need to be introduced for optometrists and ophthalmic nurses working in the public sector especially because the use of nonmydriatic retinal photography has proven to be cost-effective in the DHS of SA. 38 Automated computer analysis of digital fundus images using smartphone would be a useful tool to enhance manual analysis by various HCPs. 39 If the screening protocol is to be effectively implemented then all posts for optometrists and ophthalmologists need to be filled and posts for ophthalmic nurses in each district hospital need to be created. The continuation of the screening programme is dependent on various stakeholders at provincial and national level to incorporate this programme officially under the chronic management disease programme. This will, however, require essential resources to support the screening model.