Integrating Gestational Diabetes Screening and Care and Type 2 Diabetes Mellitus Prevention After GDM Into Community Based Primary Health Care in South Africa-Mixed Method Study

Background: Despite high gestational diabetes mellitus (GDM) prevalence in South Africa (9.1% in 2018), its screening and management are not well integrated into routine primary health care and poorly linked to post-GDM prevention of type 2 diabetes mellitus (T2DM) in South Africa’s fragmented health system. This study explored women’s, health care providers’ and experts’ experiences and perspectives on current and potential integration of GDM screening and prevention of T2DM post-GDM within routine, community-based primary health care (PHC) services in South Africa. Methods: This study drew on the Behaviour Change Wheel (BCW) framework and used a mixed method, sequential exploratory design for data collection, analysis and interpretation. Individual semi-structured interviews were conducted with key informants (n = 5) from both national and provincial levels and health care providers (n = 18) in the public health system of the Western Cape Province. Additionally, focus group discussions (FGDs) with Community Health Workers (CHWs n = 15) working with clinics in the Western Cape province. A further four FGDs and brief individual exit interviews were conducted with women with GDM (n = 35) followed-up at a tertiary hospital: Groote Schuur Hospital (GSH). Data collection with women diagnosed and treated for GDM happened between March and August 2018. Thematic analysis was the primary analytical method with some content analysis as appropriate. Statistical analysis of quantitative data from the 35 exit interview questionnaires was conducted, and correlation with qualitative variables assessed using Cramér’s V coefficient. Results: Shortage of trained staff, ill-equipped clinics, socio-economic barriers and lack of knowledge were the major reported barriers to successful integration of GDM screening and postnatal T2DM prevention. Only 43% of women reported receiving advice about all four recommendations to improve GDM and decrease T2DM risk (improve diet, reduce sugar intake, physical exercise and regularly take medication). All participants supported integrating services within routine, community-based PHC to universally screen for GDM and to prevent or delay development of T2DM after GDM. Conclusion: GDM screening and post-GDM prevention of T2DM are poorly integrated into PHC services in South Africa. Integration is desired by stakeholders (patients and providers) and may be feasible if PHC resource, training constraints and women’s socio-economic barriers are addressed.


BACKGROUND
Between 10% and 31% of type 2 diabetes (T2DM) cases among women are reported to be associated with previous gestational diabetes mellitus (GDM), and the risk of developing T2DM is increased more than 7 fold for women who had GDM compared to those without [1,2]. The national prevalence of gestational diabetes (GDM) in South Africa was estimated at 9.1% in a 2018 study [1]. The authors of the study warned of subsequent Type 2 diabetes (T2DM) for these women and their children along with complications, reduced longevity and impacts on the national health system [3].
The first step towards optimal management of GDM and prevention or delay of subsequent T2DM is diagnosis. GDM screening for all pregnant women has therefore been recommended by several professional bodies [4,5]. Currently, only a minority of women get screened worldwide for GDM [5], using many GDM testing and diagnostic criteria that have not been standardised despite efforts to do so [4] (See Table 1 for different GDM diagnostic criteria).
While necessary, screening and diagnosis alone are insufficient. Better follow up of women with GDM in order to reduce the risks of developing T2DM requires better coordination between antenatal and postnatal care [15,16]. Ideally, this could be achieved through integrated services for all conditions, or which tackle specific diseases and populations -notably the post-partum care of women's obstetric and other health care needs, and the care of infants and children. Therefore, such strategy would be feasible with an approach integrating prevention, diagnosis, treatment, and palliative care for all conditions that could be managed within PHC [17].
In South Africa, women are screened for GDM based on risk factors, as one element of ante-natal care (ANC) [18][19][20]. Women diagnosed with GDM are then referred to tertiary hospitals for their pregnancy follow-up and delivery, but only a small proportion of these women return for postpartum assessment, including an oral glucose tolerance test (OGTT), and management [21,22]. This gap between antenatal care and postnatal followup is being investigated in high income settings where many women report intentions to change their lifestyle post GDM to prevent T2DM onset, even though they find it challenging [15,23]. There is little evidence from low and middle-income countries like South Africa on actual implementation of guidelines, nor of the feasibility and acceptability of potential strategies to improve continuity and integration of care for women who have had GDM [24].
In contrast to the very low levels of routine postpartum glucose assessment following a GDM pregnancy [15], women in South Africa routinely bring their newborns and infants to clinics for immunization and well-baby care [25,26]. Integrating GDM care and prevention of T2DM post GDM within primary health care (PHC) in South Africa would facilitate women's access to services in one place. This would decrease the burden of navigating a fragmented health system for their own care and the care of their babies. In recognition of this fragmentation of care in South Africa, experts have called for integrated health systems and services that are easy for patients to navigate [27].
This study explored women's perspectives and knowledge of their own GDM and post-partum care, as well as the perspectives of health care providers, in order to assess both the current degree of integration of GDM screening and prevention of T2DM post-GDM within routine, community-based primary health care (PHC) services in South Africa. The perceived acceptability and feasibility of greater integration of these critical aspects of women's health care to these key stakeholders were also explored.

STUDY FRAMEWORK
The Behaviour Change Wheel (BCW) framework [28] ( Figure 1), was used for this study. BCW has been effectively applied to planning and evaluating interventions targeting individuals, groups, programmes and behaviours [29,30]. While motivation, capability and opportunity from the model's inner circle are seen as the enablers for both individual and collective behaviour change, they must be assessed in relation to programme implementation, its mechanism and context [31,32]. In this study, the BCW was used to understand the policies, practices and barriers to change among health workers and women.

STUDY DESIGN AND SETTINGS
A mixed method, sequential exploratory design was used, applying the Mixed Methods Appraisal Tool (MMAT), version 2018 [33,34].
This study contributes to the process evaluation of an ongoing complex intervention implementation research project, IINDIAGO (an Integrated health system Conclusion: GDM screening and post-GDM prevention of T2DM are poorly integrated into PHC services in South Africa. Integration is desired by stakeholders (patients and providers) and may be feasible if PHC resource, training constraints and women's socio-economic barriers are addressed.
Intervention aimed at reducing type 2 Diabetes risk in women after Gestational diabetes in South Africa, Trial ID: PACTR201805003336174), which aims to integrate improved post-partum follow up for women with GDM into PHC and thus contribute to T2DM prevention in two South African cities: Cape Town, Western Cape province and Soweto township in Johannesburg, Gauteng province). The IINDIAGO trial was in the recruitment stage among women from disadvantaged communities in Cape Town and Soweto, South Africa, when data were collected for this nested study. Data collection was conducted among women who attended Groote Schuur Hospital (GSH) and health care providers in health care facilities within the public health care system in the Western Cape province. Western Cape serves its population primarily through a network of clinics providing PHC services and serving as the entry point into the health care system , including two central, tertiary university teaching hospitals, an obstetrics referral hospital and one specialized pediatric hospital (Tygerberg Hospital, GSH, Mowbray Maternity Hospital and Red Cross War Memorial Children's Hospital) for specialist services [35]. PHC services in the Western Cape province are managed by two separate health authorities, Municipal City Health (in the Cape Metro Health District) and provincial Department of Health (DoH). Most district facilities are managed by the provincial DoH. The exception is Cape Metro Health District, which is managed by both City Health and provincial DoH. The Western Cape Province has 479 public PHC centres and these include clinics, of which some mobile and satellite clinics are under the authority of City Health. The provincial DoH manages Community Day Centres and Community Health Centres. All these clinics refer patients to the district provincially aided, regional, specialist and tertiary hospitals available in different parts of the province [35]. Health care providers who participated in the study were recruited from some of these clinics.

DATA SOURCES AND ELIGIBILITY CRITERIA
The data sources included: Firstly, semi-structured indepth qualitative interviews with key informants (KIs) (N = 5) and health care providers (HCPs) (N = 18). Secondly, focus group discussions (FGDs) with women diagnosed with GDM (N = 35 women in 4 FGDs) and community health workers (CHWs) (N = 15 CHWs in 2 FGDs). Additionally, exit interview questionnaires with the 35 women who participated in FGDs for further exploration of associations between qualitative variables.
The KIs included researchers, policy makers and clinicians who have been involved in DM policies and care, especially for GDM and T2DM, at national and provincial levels. They were interviewed about GDM policies and their experiences and perspectives on integrated GDM screening and T2DM prevention within PHC services. HCPs included managers, nurses or midwives from the public health sector in Cape Town (WC province), South Africa. They were interviewed about their practices or processes for GDM screening and care in facilities, including referral pathways. Drawn from these same local facilities, CHWs were recruited to FGDs to share their experiences and roles working with patients including those who had GDM and T2DM. Women with GDM referred by different clinics in Cape Town and attending GSH for their GDM follow-up and delivery, were identified from hospital records, contacted and recruited to participate in the FGDs and share their GDM knowledge and the challenges they face while seeking care. These women also completed the brief exit interviews using questionnaires, regarding their background and on how their own GDM is being managed. These datasets were used to measure the correlation between qualitative variables. All research participants discussed their views of whether and how integrated health services such as those proposed in the IINDIAGO trial could help with GDM screening and initiatives for T2DM prevention among women in SA.
Participants aged more than 18 years without any cognitive disabilities were included in this study. All were able to communicate in English. In cases of women with GDM and CHWs who did not speak English well, participants were encouraged to express themselves in isiXhosa or Afrikaans and their responses were contemporaneously translated into English by the research assistant who was fluent in these local languages. The KIs and HCPs were offered no compensation upon completion of interviews. Women with GDM who participated in this study were provided with a R100 ZAR (around $7 USD) voucher while CHWs shared refreshments after FGDs. Fieldwork and data collection were conducted between March, 2018 and August, 2018.

STUDY SAMPLE AND DATA COLLECTION
Beginning with two experts recommended by the IINDIAGO principal investigators, sequential referral snowball sampling [36] was used to identify and recruit other KIs, who were then approached and recruited for this study. Criterion sampling [37,38] was used to identify all other respondents depending on their occupations or their GDM diagnosis and referral to GSH. Managers and nurses or midwives involved in GDM screening at the clinics, CHWs who (in coordination with the local facilities) deliver services to women with different health problems in the community and assist at the clinics when called upon, were selected using this sampling strategy. Interview/FGD guides and exit interview questionnaires were respectively used as tools to collect qualitative and quantitative data. KIs were interviewed in their offices at the hospital, clinic or research facilities. The two FGDs conducted with CHWs were organised in collaboration with the two local clinics with which they were affiliated. The four patient FGDs included 6-10 women diagnosed with GDM and receiving care at GSH (N = 35), followed by individual exit interview questionnaires that took place in a room provided by the maternity ward at GSH. All interviews and FGDs were conducted by a trained researcher (JCM), assisted by a trained research assistant (SK) fluent in local languages, under the supervision of experienced qualitative researchers (KM and CZ). The researcher (JCM) introduced himself as a doctoral student and briefly interacted with the participants about the study before commencing the interviews and FGDs. Interviews with HCPs and FGDs with CHWs took place at the clinics, in their clinic offices for HCPs and in the rooms provided by the local clinics for CHWs. Each interview lasted between 30 and 45 minutes. FGDs lasted between 45 minutes and 1 hour. The Exit interviews lasted between 10 and 15 minutes. All interviews and FGDs were audio recorded and ATLAS.ti software was used to assist data analysis and management.

DATA ANALYSIS
The interviews and FGDs were transcribed and a coding system was developed by JCM in collaboration with CZ using an inductive/deductive approach. All discrepancies in the coding process were discussed and resolved between these two investigators. Thematic analysis was generally used but content analysis was applied on a few occasions in order to check the frequency of important codes [39,40]. For statistical analysis of the 35 exit interviews questionnaires, categorical variables were summarized using absolute frequencies and relative frequencies. Continuous variables were synthesized using central trend statistics (mean, median) and dispersion statistics (standard deviation (SD), interquartile range (IQR)). Qualitative variables were four advices for women (improve diet, reduce sugar intake, physical exercise and regularly take prescribed medication) to improve their GDM and prevent T2DM and Nurse's concerns about health of these women. The correlation between these qualitative variables measured using Cramér's V coefficient which is interpreted as follows: from 0.0 to <0.1 negligible association, from ≥0.1 to <0.3 weak association, from ≥0.3 to <0.5 moderate association and ≥0.5 strong association [41].
This analysis has also contributed to the ongoing process evaluation of the IINDIAGO study.

RESULTS
In total, 73 individuals participated in this study. Participants in the in-depth individual interviews (N = 23), included 4 (17%) clinic managers and 14 (61%) nurses and midwives and 5 expert KIs (22%). Of these 23 respondents, 19 (83%) were female, with a mean age (SD) of 42.7 (SD 10.6) years and 16.1 (SD 11.0) years of experience in health care (see Table 2). Participants in FGDs (N = 50) included women with GDM and CHWs and were all female.
The four thematic categories that emerged from the analysed data were interpreted using three BCW layers from outer to inner: policy categories, intervention functions and sources of behaviour respectively. Each category was linked to a specific layer except the third and the fourth categories that were classified using the same "inner" layer (See Table 3).
Each category had different themes with each illustrated by a single quote from one of the participant groups. More illustrative quotes from various participants are depicted in Table 4.

THE EXISTING GUIDELINES, SERVICES AND CURRENT PRACTICES IN THE CLINICS
The BCW's policy categories or outer layer [31] was used to assess the process of policy development, analyse its implementation and interpret insights from the KIs and  the HCPs regarding the existing guidelines, services and current practices in the clinics regarding GDM screening, care and post-partum T2DM prevention initiatives. Perspectives and experiences of CHWs and women who participated in FGDs mostly referred to services they received and the practices in the clinics they attended.
The results for this thematic category were subsequently grouped into 3 headings: 1. Current GDM screening/care guidelines and their implementation; 2. From no testing to risk-based screening of GDM; and 3. Barriers to GDM screening in PHC.

CURRENT GDM SCREENING/ CARE GUIDELINES AND THEIR IMPLEMENTATION
Both National and Western Cape departments of health introduced guidelines developed by experts based on international protocols to screen, diagnose and treat GDM [21], in all public health facilities. However, challenges arise in the implementation of these guidelines at local facilities. While discussing how GDM is diagnosed within ANC, KIs and HCPs reported that GDM screening guidelines have been poorly implemented at primary care level, resulting in missing some women with potential GDM.

PROCESS OF GDM SCREENING AND REFERRING WOMEN DIAGNOSED WITH GDM
The process of diabetes screening during pregnancy based on current guidelines in the local facilities in Cape Town is summarized in Figure 2. Referral starts from BANC, to Midwife and Obstetrics Units (MOU), to secondary level specialised maternity hospitals in case of impaired glucose tolerance (IGT -in which plasma glucose levels were above normal but below those defined as diabetes) [42,43], to tertiary hospitals (GSH or Tygerberg hospital depending on jurisdiction of the MOU) for cases meeting local criteria for GDM.

Current GDM screening/care guidelines and its poor implementation
"So, what we basically do in our facility, so we go according to the BANC protocol. We have our own protocol. If a mother comes in the morning for an antenatal booking, then we test her urine…". HCP 1.
2. From no testing to the risk-based screening of GDM at the clinics i. Risk factors assessment "There are two Community Centres in Gugulethu, the, and then it's us, the mobile Baby Clinic. In our clinic it's basic antenatal care, so the people who have a history with parents who are diabetic, usually we send them to the MOU, they are screened that side. We don't do screening in our clinic. We don't actually do that". HCP 3.
"Not every mum, but if she presents risk factors such as a family history of diabetes, the mum had a previous history with Gestational Diabetes, she has an exceeding Body Mass Index (BMI) and then if we tested the random blood sugar and found that it was above 7.8, then we will give the mother a Gestational Diabetes check…". HCP 4.
"So we do a random blood glucose at the facility, and depending on that result, we will then follow the necessary steps. There is obviously a screening in terms of family history, and have you had Diabetes before, or do you currently have Diabetes….". KI 6.
ii. Process of GDM testing and referring women with GDM "the procedure for screening, we've got a list of indications for doing Glucose Tolerance Test (GTT): family history of diabetes from her mother, her father or her siblings, BMI of 35 and above, history of big babies, persistent Glycosuria; for three consecutive visits. She has to come in the morning, fasting, her last meal the previous night around 10 o'clock. So, when she comes, we do the prick. If the sugar is 7 and above, we don't continue, but if it is less than 7, we take the fasting blood and we give her 75 grams of glucose, and we take the second blood after two hours. So, they come after one week for the results. If it's an IGT, we refer to Mowbray not Groote Schuur, but if it's GDM, then we refer to Groote Schuur". HCP 2.
"when they come here for the first time, we do the IGT (Impaired Glucose Tolerance) test or sugar test, and then if there is family history like the mother was diabetic, then we do the OGTT test, which is the fasting glucose, but we don't do it here. I have to book for them in Gugulethu, and then they are going to give me the date when the patient can go there. Otherwise we have the forms that we use. We just take… I'm going to show you later the forms, and then we take, if the patient has already diabetes and she does not qualify to book here at the clinic, so I refer the patient straight to Gugulethu MOU". HCP 5.

Barriers to GDM screening into PHC
"Well, the current practice is to try and identify them from women who attend antenatal care. That obviously means, the people who don't attend, we wouldn't pick it up…" KI 2.
"You know, unfortunately a lot of the patients are picked up a bit later. The patients we pick up earlier of course, are those who previously diabetes, which is a different ballgame. So those get to come in early, but the majority of the patients come in at a later time…". KI 3.
"The only challenge is that when you give an appointment for the lady to come to do bloods, then she doesn't come. Then it will be picked up because they are supposed to do it before they are 28 weeks; or if you do it at 28weeks then you have to repeat it. If it was borderline then you have to repeat, so then you don't have that chance of checking if you pricked them already at seven months or close to eight months, so you don't have that chance of checking, then you are going to refer them, because they are already late in pregnancy". HCP 6. "We don't have time to talk individually, but at times when we give the Health Talk, we do explain to them…". HCP 4.
"You call an ambulance to pick up the clients to take to the MOU, or Mowbray, depending on where the pathway is. Now we send the letter. On the letter there is a sleeve that is supposed to come back to us, but that has never happened. I have been here for more than eight years now, I have never seen that sleeve coming back...". HCP 7.
"She must bring her own food, because we do not have glucose to eat. She goes and has breakfast, and then two hours later we re-prick…". HCP 1.

Category 2: Effective antenatal referral procedures but lack of follow-up after delivery
i On-site integrated hospital services "she gets referred to Groote Schuur Hospital's antenatal clinic where they will do what we call OPD (Out-Patients Department) spreads, and then they will start treatment; but the first line treatment for any diabetic is diet, and so she will see the dietician, lifestyle changes, and then she will start treatment…". HCP 8.
"nurses play an indispensable role in managing these patients, bearing in mind that the maternal and foetal wellbeing will be first assessed by nurses, and also nurse will also help in providing anthropometric measurements, they help to also reduce the time-lapse in some of these patients to spend a very long time waiting for doctors. So basically, nurses play a role in monitoring of the mother and the baby, as well as even sometimes in diagnosis and also in management". KI 5.

(Contd.)
We all have our specialities, so the registrar that would be looking after the patient is somebody that is rotated through the whole block, so they've seen cardiac, they've seen eclamptic patients, they've done diabetes; but if there is a specific problem, then we are in the fortunate position where we have the resources where we can get infectious disease people out, instead of struggling with that, or we can get the endocrinologist out, and say listen, we have now hit a wall, how do we go forward, but that is within our setting". HCP 8.
ii Socio-economic boundaries to healthy antenatal and postnatal initiatives "…. sometimes when you check in, they say you must come without eating to the clinic and then they take a long time to check your sugars, and then you get tired, you are hungry. You know how you are when you're hungry, you seriously want to". Participant in FGD 1.  "So, I think firstly the doctors and nurses don't always have enough time, and also, they're not very knowledgeable, and then even the dieticians are sometimes giving the wrong messages because of this whole debate internationally. So, I think those things are a problem, and then there is also the issue of healthy foods being expensive in townships, and the issue of exercise is difficult. I mean, if women get up at five o'clock, go to work in the town, go back, don't get home till seven, you know, their lifestyles aren't conducive to exercise". KI 4.
"The maternity sisters do not communicate with the local clinic sister for follow-up on these clients about medication after delivery and then we don't know. So, maybe they got letters from hospital that you must follow up at this clinic to get your medication that is going to control you but mothers don't follow up, as I have noted, they don't follow up, they only focus on the baby after delivery, they focus on the baby. They don't go for that follow up appointment and the medication, especially after they are coming from Maternity. But if there is a problem, then the doctor prescribes when discharging them but they will never mention it to us at the clinic…And then, if they are with the person who didn't see them when pregnant, you won't know if the client had a problem with the glucose". HCP 10.

"It is now policy. We had it two or three years ago, we wrote a postnatal care policy for the Western Cape, and I was involved in writing it, and it's agreed, it's just no-one has implemented it. So, it has to be implemented.… So, I do think it needs to be resourced. You need another nurse, and you need a particular training to give that nurse the referral route. So, what does she do with a person who's depressed at six weeks? What does she do with the one who had GDM and they've checked her sugar now and it's normal? What do they do with her? So, I think it needs almost a little bit of a syllabus for what the nurse does, you know?". KI 4.
We see it with IGT patients who are very well counselled and can actually reverse the whole and become normal. So, I think it's feasible. I think it's good that it (IINDIAGO) will give you raw data that you can then present to policymakers and say, listen, although we knew this, this is the hard data, done in a methodologically robust manner, and that no-one can argue with. And once faced with that, then one will have to change policies, and be forced to change the infrastructure and the way the infrastructure is set up to deal, not only with Type 2 Diabetes but with many other problems. KI 3.
"I think if we can implement it (IINDIAGO) at the Well Baby Clinics for instance if they have enough staff and they are well-trained, I think it would make a big difference, because as a mother sometimes you are more worried about your baby, so then you are more likely to access that service; and then I think, like I said earlier, a continuation of care is better… So, if she has that continuous support at the Well Baby Clinic, because that is a place where she will be accessing the services quit e frequently, so she will be able to build a bond or a type of relationship with that caregiver on that side as well". HCP 8.

"It (IINDIAGO) is a good thing, because we such type of intervention we will normally check if everything is good when they come for post-delivery.
We now just focus on breastfeeding and not in that side. We don't go on the Diabetes side and Hypertension and all those things". HCP 6.

"You see, I think because the women with GDM after pregnancy, most of them don't have any medical problem, the doctor or nurse will think they are wasting their time at the clinic, so, actually the initiative should be a community based one through lifestyle, and I think the community health workers are most important…but I don't know the answers about the scope of a community health worker". KI 4.
"We also have health care workers that are not based in the clinic, but they report. Those are the people that are helping us work or supervising the ART or TB treatment for the people that are placed in the community to take their treatment. They visit. Even with the immunisation that is really not doing so well, they are able to the visits, the home visits. They are in contact with the community, so they also can help in this intervention (IINDIAGO)". HCP 10.
"…they also help us with recalling the mothers for other things. I think they can also play a role in this intervention (IINDIAGO)". HCP 11.
"For me, I think that the community workers are people from the community, so, the patients trust them more than coming to a sister in a hospital they can only see once.… so, the community knows them. If they do the screens and stuff they tend to trust them more than us some times. Yes, I think they need to be trained, because the last time I said they even need to be trained in doing prognostics for us, then they can do the diabetic screening at the same time, different screenings; because they are there in the house with ten people around them, so they can do all of that, and then they catch them early, even the blood finger prick". HCP 4.

"Me, I love the job that I am doing because I don't have a problem with people, and I can convince them but if someone is not doing well, I report her to the supervisor who will then intervene". Participant 3, FGD 1.
"My challenge is work load. We have to record. We have to be ready to give weekly and monthly statistics for our work. It's a challenge, because there is a lot of work. We have to visit the clients, rain or shine, you have to visit them. You must have the minimum six to eight, and then each and every day you must have something to write down as proof of what you have done for the day. We must also cover many households at a long distance and reach target…. Participant 5, FGD 1.
"Sometimes when we arrive at a patient, we see a number of men smoking. The whole house is like snow, so I am afraid of entering that house fearing what could happen to me when I enter that house". Participant 3, FGD 2.
"Challenges also include robberies in the community and even here at the clinic, they just come and attack you at the clinic's gate and sometimes we are not working with our cell phones because we are afraid of robbery by the gangsters. And sometimes, even in the houses that they are going to do the pill counts in, they mustn't go alone. We must therefore be two or three but it is not easy to get that one to make a friend and go together to avoid those incidents". Participant 4, FGD 2. Table 4 Categories and illustrative quotes.

BARRIERS TO GDM SCREENING IN PHC
Respondents identified several barriers to screening. First of all, socio-economic factors impede the timely access of many women to BANC and to GDM screening. Secondly, some may be diagnosed late when symptoms or consequences of GDM are already present; this becomes a reason for immediate referral to the hospital for follow-up. Thirdly, there is no way to identify and screen some women as they do not attend ANC at the clinics at all. Fourthly, as many women, like other patients, do not know or suspect that they might have diabetes, they do not proactively seek any screening during their pregnancy or clinic visits, which is why the provincial guidelines mandate that screening be initiated by the provider. Finally, lack of time due to work overload, shortage or ineffective utilisation of key equipment and other resources for GDM screening, and poor communication between facilities, were also included among other documented health system issues preventing consistent GDM screening in PHC.
"At the moment I think we have got one glucometer in the whole clinic, you understand? Sometimes we don't know where it is and it is difficult to find it, you see…". HCP 2.

EFFECTIVE ANTENATAL REFERRAL PROCEDURES BUT LACK OF FOLLOW-UP AFTER DELIVERY
Themes in this category were appraised in the light of BCW's intervention functions or middle layer [31], regarding services offered to women diagnosed with GDM while attending a diabetic clinic at hospital for follow-up and delivery.

ON-SITE INTEGRATED HOSPITAL SERVICES
Upon arrival at the respective hospital to which they are referred, women with GDM benefit from hospital level integrated care under the coordination of the diabetic clinic of the maternity department. Integrated services at referral hospitals include regular blood glucose monitoring, investigations for other health problems, medical care for GDM and other health problems beyond GDM, as well as diet and lifestyle change interventions.
"We all have our specialities, so the registrar that would be looking after the patient is somebody that is rotated through the whole block, so they've seen cardiac, they've seen eclamptic patients, they've done diabetes; but if there is a specific problem, then we are in the fortunate position where we have the resources where we can get infectious disease people out, instead of struggling with that, or we can get the endocrinologist out, and say listen, we have now hit a wall, how do we go forward, but that is within our setting". HCP 3.
Counselling sessions regarding lifestyle changes to deal with diabetes and its devastating consequences were said to be routinely scheduled but not integrated within the services offered in the diabetic clinic of GSH's obstetrics unit. As noted above, respondents reported that when counselling sessions were offered, crowded clinic conditions and lack of privacy decreased the effectiveness of sessions.

SOCIO-ECONOMIC BARRIERS TO CHALLENGES TO HEALTHY ANTENATAL AND POSTNATAL INITIATIVES
Women diagnosed with GDM at primary care and referred to tertiary hospital (GSH) (N = 35) discussed barriers they faced in their long road to care from families/ communities, local facilities and up to referral hospitals, with many visits both during and after pregnancy. For many women the transport costs to attend care and the extra cost of healthy food contributed to depleting their already constrained economic resource.

CONFUSION OR LITTLE KNOWLEDGE ABOUT GDM AND LIFESTYLE CHANGES
Apart from the socio-economic issues that women have to deal with in their daily lives, many have shown confusion or limited understanding of what GDM is and the behavioural/lifestyle changes required to manage GDM and prevent or delay future T2DM for themselves and long-term metabolic problems for their babies. Some women could not explain clearly what GDM was or why a particular treatment was prescribed to them while others struggled to name GDM consequences for themselves and their babies.
"I also think GDM is when you are diabetic, they find out when you are pregnant, and then it's not going to be seen after birth, but I was thinking like that before, but I have never actually known…that's an impression, but I'm just assuming, I'm not sure".

Participant 1 in FGD 4.
Women's understanding about GDM as discussed in the FGDs was compared to the results from exit interview questionnaires. Despite the time they spent throughout the diagnosis and referral process at lower levels of health care, and after attending the diabetic clinic at GSH for their GDM care many times, only 43% reported having received advice about all four recommended actions (improve diet, reduce sugar intake, physical exercise and regularly take prescribed medication) to improve their GDM and prevent T2DM. However, women reported being satisfied with the information they had received, despite this lack of alignment with recommendations. Only half (51%) of the respondents were aware of the importance of reducing sugar intake, while 69% recalled being advised to exercise, 86% to improve their diet and 83% to take pills regularly. The contrast of improving diet (86%) and reducing sugar intake (51%) suggests incomplete and/or ineffective lifestyle change education. Table 5 shows that, in contrast to the barriers reported by respondents regarding selective and late screening practices at primary care level, 94% kept their appointments at the referral hospital (GSH), 49% had already been tested in the morning before the FGDs were conducted, and 77% felt that nurses were interested or concerned about their health. Women who felt that nurses have empathy and time for them easily engaged with the nurses to ask about their GDM and general health, trusting their advice to change their lifestyle during pregnancy and postpartum to prevent or delay T2DM onset. Most women had their appointments every week (43%) or every 2 weeks (34%) with 86% reporting having received all their medications and not facing any stock-out. The multiple correlations between the advice that women with GDM received and their view on whether nurses were interested or concerned about their health, generally established a negligible, weak or moderate association as none reached 0.5 ( Figure 3).

POOR COMMUNICATION AND NON-EXISTENT PLANS FOR POSTNATAL FOLLOW-UP
Even though integrated services including ANC and GDM were generally appreciated at referral hospitals, there was no follow-up for women and their babies after delivery. When women and their babies were back in their community after delivery, it was reported that they were seen at their local facilities exclusively for babies' check-up and immunisation and that women did not have access to any specific programme that provided follow-up. Upon discharge, the details about their health were written up in their antenatal record (Road to Health booklet), but according to women and HCPs, the hospital does not consistently give or send a referral letter, nor call or communicate in other ways with local clinics regarding postnatal care. This is despite the guidelines indicating that a referral letter recommending a 6 week postpartum OGTT and follow up at a local clinic should always be handed to women at discharge. Both KIs and HCPs reported that, once back at the clinics, the women who did receive follow-up letters and medication from hospital tend to focus on their babies and forget or ignore to look after their own health. The few women who had approached the nurses at the primary care clinic regarding their postnatal check-up reported that they had not been successful in getting screened for diabetes.
"The maternity sisters do not communicate with the local clinic sister for follow-up on these clients about medication after delivery and then we don't know. So, maybe they got letters from hospital that you must follow up at this clinic to get your medication that is going to control you but mothers don't follow up, as I have noted, they don't follow up, they only focus on the baby after delivery,…... But if there is a problem, then the doctor prescribes when discharging them but they will never mention it to us at the clinic…And then, if they are with the person who didn't see them when pregnant, you won't know if the client had a problem with the glucose". HCP 3.

IINDIAGO, AN INTERVENTION WITH POTENTIAL TO BRIDGE THE GAPS IN POSTNATAL FOLLOW-UP
The BCW's sources of behaviour or inner layer [31] was used to map and interpret failures in postnatal followup for mothers with previous GDM and their babies, a problem identified by all KI and HCPs approached for this research. This reported gap in postnatal care for women with GDM was also seen as an implementation vacuum that the newly approved WC postnatal policy aimed to solve [44] but respondents considered that the policy fell short in terms of follow-up for women who had GDM. The ongoing IINDIAGO study that aims to integrate postpartum follow up for women post-GDM into PHC was presented to respondents at the end of the interview, in order to explore the perceived relevance and feasibility of such an intervention. The idea was welcomed and seen as feasible by all respondents including women.

APPRECIATION OF CHWS INVOLVEMENT IN COMMUNITY BASED T2DM PREVENTION INTERVENTION
The BCW's sources of behaviour or inner layer [31] was applied to help understand the changes that need to take place in the community in order to prevent or delay   T2DM. Since the overwhelmed clinics do not intervene much, if at all, in T2DM prevention efforts, CHWs were considered to be the best-placed health workers to successfully contribute to implementation of activities in the family and community. CHWs in South Africa have greatly assisted [45,46] in other communitybased interventions to improve health, principally in the areas of maternal and child health and HIV care. Existing policy also gives them a role to play in noncommunicable diseases (NCDs) [47]. All participants (KIs, HCPs, Women) commended CHWs and suggested that they get involved in T2DM prevention once trained and working under clinic supervision. KIs suggested that their involvement could bring some clinical services like NCDs screening, counselling, health education, and implementation of specific preventive measures to the patients and family members within communities.
Reflecting on their experiences with HIV and tuberculosis, the CHWs who participated in this study responded positively to the idea of getting involved in such innovative and integrated approach towards T2DM prevention for women who had GDM. CHWs explained how their visits to the families within community are more inclusive and go beyond the single patient they are scheduled to visit, covering a range of health problems of all present family members. Equipped with their household charts, they reported that they conduct a complete surveillance of the family and refer family members with particular health problems to the right health facility for further diagnosis and care. CHWs emphasised their visiting and educating roles would align well with the tasks they would handle in T2DM prevention efforts. These positive comments were made despite reporting challenges they face in their daily activities like limited training; low and irregular payments; very busy clinics that sometimes fail to follow up the patients they refer to them. CHWs expressed commitment to their cause and engagement in their mission within the community.
"Me, I love the job that I am doing because I don't have a problem with people, and I can convince them but if someone is not doing well, I report her to the supervisor who will then intervene". CHW participant 3, FGD 1.

DISCUSSION
In the face of increasing GDM prevalence in Africa [48] and despite calls for universal screening, the guidelines in most countries recommend selective screening to diagnose and manage GDM and its sequelae [49,50]. Risk factor-based screening has been the main approach adopted in South Africa. Even though the current GDM screening guidelines in South Africa now meet international standards, respecting the value thresholds as recently discussed by Adam S. and Rheeder [19], they are still ineffectively applied. The "Basic antenatal care (BANC) protocol" was identified as the main tool used for antenatal service provision in most of Cape Town clinics but is a complex guideline with many components [17].
Ultimately the decision to screen GDM or not lies with the nurses, in line with the facility plan rather than this complex protocol itself. Documented challenges in GDM testing at primary care level were a sign but also a cause of poor screening practice. Universal screening of GDM cannot be successful if concurrent barriers are not addressed. These challenges to GDM screening in PHC include but are not limited to shortages of well-trained HCPs and ill-equipped clinics to test and deal with NCDs based on the available guidelines [51]. Multiple barriers impeding proper GDM screening and follow up post-GDM have been documented in other studies and this study's findings corroborate many, including: weaknesses at different health system levels; poor understanding of postpartum GDM risks of T2DM development for both women and their babies; and various patient, community and health service level barriers for women when they are referred back into PHC for follow-up after delivery [15,52,53]. Our findings further suggest that the expertise and knowledge required of both nurses and women are insufficient to make a risk factor-based approach effective in South Africa.
It has never been easy for women to navigate health systems to access obstetric care in sub-Saharan Africa due to multiple individual and family socio-economic barriers such as low household income, illiteracy, lack of transport means and its cost, and cultural beliefs/ practices, among others as reported in recent studies [54,55]. Despite these issues, women receive integrated and highly appreciated antenatal and perinatal care at the tertiary level. Women with GDM who participated in this study confirmed this. However, women who strived to protect their babies from the adverse effects of GDM feel relieved after delivery and this is reinforced after their glucose levels return to the normal range. Additionally, the lack of structured postnatal care for these women does not foster the implementation of T2DM prevention initiatives.
Our findings suggest that this could be at least partially mitigated with clear and consistent discussions about GDM and its long-term consequences for both women and their babies throughout ANC, perinatal and postpartum services. Health education may encourage these women to follow-up with postnatal testing and lifestyle change measures at the clinic and in the community. Referral hospitals must first communicate with the local facilities regarding follow-up for these women and, in return, the clinics need to continue surveillance and initiate integrated postnatal behavioural change interventions for T2DM prevention. Such interventions would be useful for other NCDs and broader health care needs beyond the immediate aim of dealing with IGT, T2DM or diabetes related health issues but to achieve this, nurses need appropriate training and more resources in the facilities.
The IINDIAGO project is exploring whether such postnatal follow-up could be linked to the babies' immunisation, which normally starts soon after delivery and discharge from hospital. HCPs showed willingness to add this programme to their workload after receiving proper guidelines and adequate training on their side. Women also expressed support for this kind of intervention after discussing its dual benefits, for them and for their babies. Engaging policy makers to change guidelines on the one hand and appropriately train frontline healthcare workers including CHWs on the other has succeeded in other trials and interventions in PHC for the same populations. Here, Prevention of Mother-to-Child Transmission of HIV (PMTCT) which continues from ANC into postnatal care with lifelong services within the facility and in the community [56][57][58] could serve as a case study.

CONCLUSION
Effective care of GDM and prevention or delay of T2DM requires a continuum of care from screening and diagnosis of GDM, to antenatal and intrapartum management, to post-partum follow up and prevention interventions. Despite policy support and guidelines promoting integrated care, implementation of GDM screening, delivery of counselling about GDM and T2DM, and post-partum follow up are suboptimal in Western Cape. Many women are diagnosed late in their pregnancy and postnatal follow-up is almost nonexistent. An innovative strategy of integrating universal GDM screening in local health facilities with postnatal follow-up of these women and their babies in the community based PHC services is considered desirable and feasible by all participants in this study. Women, health providers, and experts added that this integration would work well if the resource and training constraints facing PHC as well as socio-economic barriers to women are addressed.

DATA ACCESSIBILITY STATEMENTS
The datasets analysed during the current study are not publicly available to preserve participant anonymity.

ADDITIONAL FILE
The additional file for this article can be found as follows: