Improving systems of care during and after a pregnancy complicated by hyperglycaemia: A protocol for a complex health systems intervention

Background Many women with hyperglycaemia in pregnancy do not receive care during and after pregnancy according to standards recommended in international guidelines. The burden of hyperglycaemia in pregnancy falls disproportionately upon Indigenous peoples worldwide, including Aboriginal and Torres Strait Islander women in Australia. The remote and regional Australian context poses additional barriers to delivering healthcare, including high staff turnover and a socially disadvantaged population with a high prevalence of diabetes. Methods A complex health systems intervention to improve care for women during and after a pregnancy complicated by hyperglycaemia will be implemented in remote and regional Australia (the Northern Territory and Far North Queensland). The Theoretical Domains Framework was used during formative work with stakeholders to identify intervention components: (1) increasing workforce capacity, skills and knowledge and improving health literacy of health professionals and women; (2) improving access to healthcare through culturally and clinically appropriate pathways; (3) improving information management and communication; (4) enhancing policies and guidelines; (5) embedding use of a clinical register as a quality improvement tool. The intervention will be evaluated utilising the RE-AIM framework at two timepoints: firstly, a qualitative interim evaluation involving interviews with stakeholders (health professionals, champions and project implementers); and subsequently a mixed-methods final evaluation of outcomes and processes: interviews with stakeholders; survey of health professionals; an audit of electronic health records and clinical register; and a review of operational documents. Outcome measures include changes between pre- and post-intervention in: proportion of high risk women receiving recommended glucose screening in early pregnancy; diabetes-related birth outcomes; proportion of women receiving recommended postpartum care including glucose testing; health practitioner confidence in providing care, knowledge and use of relevant guidelines and referral pathways, and perception of care coordination and communication systems; changes to health systems including referral pathways and clinical guidelines. Discussion This study will provide insights into the impact of health systems changes in improving care for women with hyperglycaemia during and after pregnancy in a challenging setting. It will also provide detailed information on process measures in the implementation of such health system changes.


Background
Hyperglycaemia in pregnancy, encompassing gestational diabetes mellitus (GDM), pre-existing type 2 diabetes (T2DM) and overt (likely type 2) diabetes in pregnancy, is associated with adverse health outcomes for mother and child, both in the peripartum period and long-term [1,2]. The International Federation of Gynecology and Obstetrics (FIGO) recently identified improving systems of care for women with hyperglycaemia in pregnancy, particularly during the postpartum period, as a research priority [3]. The period during and after pregnancy is an opportune time to optimise maternal health, which is in turn an important strategy to reduce the risk of adverse outcomes in any future pregnancy. International guidelines provide recommendations for the care of women with hyperglycaemia in pregnancy, including postpartum glucose screening and counselling regarding contraception [4,5]. However, there are significant gaps in implementing care which meets these recommendations, with a staggering 75-80% of women lost to follow-up [3], and an average of only 33% of women internationally completing postpartum glucose testing following GDM [6].
An estimated 16.9%, or 21.4 million, live births around the world are complicated by hyperglycaemia each year [7]. This burden falls disproportionately upon Indigenous women globally [8,9]. In Australia, Aboriginal and Torres Strait Islander women are 10 times more likely to have pre-existing T2DM in pregnancy, and 1.5 times more likely to develop GDM [10,11]. GDM is a strong predictor of future T2DM [12]; the risk of progressing to T2DM following GDM is fourfold greater for Aboriginal and Torres Strait Islander women than non-Indigenous women in Australia [13]. Children exposed to hyperglycaemia in utero are at risk of developing T2DM at an early age, which is an issue of increasing concern for the health of Aboriginal and Torres Strait Islander Australians [14,15]. The consequences of hyperglycaemia in pregnancy contribute substantially to the epidemic of diabetes, and thus to the 10-year gap in life expectancy between Aboriginal and Torres Strait Islander peoples and non-Indigenous Australians [16,17]. To address disparities in health outcomes between Aboriginal and Torres Strait Islander and non-Indigenous Australians, there is an urgent need to reduce risk as early as possible in the life course.
Prior interventions to improve care for women with hyperglycaemia in pregnancy have predominantly focussed on increasing the proportion of women attending glucose screening after GDM. Single component interventions, such as patient education programs, postpartum reminder systems or use of checklists, have not consistently demonstrated improvements [18][19][20][21]. This contrasts with multi-component interventions, which have achieved postpartum glucose screening rates as high as 95.8% [22][23][24][25]. Measures utilised in multicomponent interventions include instituting protocol-based care, reminder systems, introduction of a nurse navigator, and education for healthcare providers and women. The impact of such intervention components is yet to be demonstrated in the regional and remote Australian context, where there are multiple barriers to implementing guideline-based recommendations for the care of women with hyperglycaemia in pregnancy. These barriers include a disproportionate burden of socioeconomic disadvantage, population mobility, geographic remoteness, high turnover of clinical staff, and fragmentation between service providers [26].
The Northern Territory Diabetes in Pregnancy (DIP) Partnership formed in 2011 between health service providers, policymakers and researchers to improve the care of women with hyperglycaemia in pregnancy. The Partnership expanded in 2016 to include the region of Far North Queensland, and more recently to consider the intergenerational impact of diabetes, and thus is now the "Diabetes Across the Lifecourse: Northern Australia Partnership" ("the Partnership"). Previous work of the Partnership has included improvements in antenatal service delivery for women with hyperglycaemia in pregnancy in the Northern Territory [26] and the establishment of the DIP Clinical Register [27].
Building on this previous work we have developed a multi-component health systems intervention to improve care for women across regions of northern Australia during and after a pregnancy complicated by hyperglycaemia. Components, which include reminder systems and health practitioner education, have been selected based on evidence for improving care of women with hyperglycaemia in pregnancy in other contexts [23,25]. An additional component of our intervention is use of the DIP Clinical Register as a recall and continuous quality improvement tool. Clinical registers have been widely utilised, including in regional and remote Australia, to improve systems of care, clinical follow-up and health outcomes for chronic conditions [28,29]. The Partnership has undertaken formative research with health professionals and stakeholders to identify gaps in care, which contributed to further refining our intervention design [26,30,31].
This protocol describes the planned implementation of our health systems intervention, in accordance with the Template for Intervention Description and Replication (TIDieR) [32] and the Revised Standards for QUality Improvement Reporting Excellence (SQUIRE 2.0) (Supplementary Materials) [33]. These frameworks both provide guidance intending to improve the completeness of reporting, and thus replicability, of healthcare interventions, with SQUIRE 2.0 specifically focussing on health service improvement.

Aim
To develop, implement and evaluate a health systems intervention to improve models of antenatal and postpartum care for women with hyperglycaemia in pregnancy in regional and remote Australia.

Design
This study will use a cross-sectional before-and-after design to assess the impacts of the health systems intervention.

Theoretical Framework
Formative work was conducted with health professionals using the Theoretical Domains Framework (TDF) [34], identifying barriers to the implementation of care according to local guidelines [35,36] for women during and after a pregnancy complicated by diabetes in remote and regional Australia. The TDF describes 12 domains covering the main factors that influence health practitioner clinical behaviour and behaviour changes: knowledge; skills; social/professional role and identity; beliefs about capabilities; beliefs about consequences; motivation and goals; memory, attention and decision processes; environmental context and resources; social influences; emotion; behavioural regulation; and nature of the behaviours. Use of these domains enables identification of a wide range of potential barriers to implementation of improvements in health systems, facilitating development of multiple potential intervention components to overcome the barriers identified.
Detailed barriers and enablers identified by this formative work have been previously reported [31,37]. In brief, multiple factors impacting on health service delivery were identified, including fragmentation of care, gaps in communication and a lack of clarity regarding healthcare provider responsibilities for components of care such as screening and post-partum follow-up [37]. Further details regarding major barriers are described in Table 1. Opportunities to improve care were also identified, including enhancing education and support for health professionals and improving communication pathways [37].

Setting
The Northern Territory (NT) and Far North Queensland (FNQ) encompass a geographic area of over 1.6 million square kilometres, including numerous remote islands. The region is sparsely populated, with approximately 500,000 inhabitants [38][39][40]. Approximately 22.5% of the population identifies as Aboriginal and/or Torres Strait Islander peoples, compared with 3.2% across Australia [38][39][40][41]. There is a high degree of cultural diversity, with over 200 languages spoken [40]. There are approximately 7000 births across the region annually [11,38,39]. In the NT in 2015, 11.9% of births to all women were complicated by GDM and 1.7% by preexisting diabetes; for Aboriginal women, these rates were 15.4% and 4.1% respectively [11]. While the official numbers of births affected by diabetes in FNQ are not publicly reported, an audit of births to Aboriginal and Torres Strait Islander women showed the prevalence of GDM and T2DM to be 14.2% and 2.3% respectively [42].
This health systems intervention will be conducted across tertiary, secondary and primary healthcare centres, including government and Aboriginal communitycontrolled organisations, throughout three study regions (Central Australia and Top End, within the NT; and FNQ) ( Figure 1).

Participants
Health professionals across the NT and FNQ who are involved in the care or care coordination of women with hyperglycaemia in pregnancy.

Procedures
Formative focus groups conducted with health professionals in 2016-17 led to the identification of five key models of care components: 1. Increasing workforce capacity, skills and knowledge and improving the health literacy of health professionals and women. 2. Improving access to healthcare through culturally and clinically appropriate pathways. 3. Improving information management and communication.
4. Enhancing policies and guidelines.

Embedding the NT and FNQ Diabetes in Pregnancy
Clinical Register [27] within the models of care as a continuous quality improvement tool.

Implementation Activities
The above components have guided the development of implementation activities with local healthcare professionals and stakeholders to address identified barriers (Table 2; Figure 2). A project coordinator will lead implementation of these activities in each study region, with support from the central project office. Activities will be implemented across primary, secondary and tertiary level services throughout each study region, aiming to reach all health services across the regions providing care to women with hyperglycaemia in pregnancy. Due to differences in health systems between the study regions, it is anticipated that there will be context-specific variations in implementation activities across regions. Delivery of activities will be recorded by implementation staff in an Activity Log, including nature of activity, study region, Models of Care component addressed, number of attendees (where relevant), involvement of other organisations and feedback. Activities and associated materials will be modified throughout the delivery of the project based on feedback from health professionals, project implementers and other stakeholders. The interim evaluation (see below) will provide the main opportunity for feedback and modifications; additional opportunities will be facilitated through regular meetings with stakeholder groups, including a Clinical Reference Group and Indigenous Table 1 Barriers to implementing healthcare according to local guidelines for women during and after a pregnancy complicated by diabetes in regional and remote Australia, identified in formative work with healthcare professionals and stakeholders Reference Group in the NT, and a Partnership Working Group in FNQ. Key implementation activities can be grouped into: educational sessions and resources; recall and reminder systems to assist with follow-up; establishing (in FNQ) and expanding (in NT) the use of the Diabetes in Pregnancy Clinical Register; and stakeholder engagement and consultation ( Figure 2). There is substantial overlap between these activities, and many activities align with several of the Models of Care components. Further detail is provided in Table 2.
Educational Sessions and Resources Implementation staff will collaborate with expert clinicians in the development and delivery of education to health professionals involved in the antenatal and postpartum care of women with hyperglycaemia in pregnancy. Face-to-face and online educational sessions will be designed to reach relevant healthcare providers, with sessions promoted through professional networks and health services. Education will promote evidence-based care according to local guidelines. Educational materials will be made available to interested health professionals for distribution to networks. Key messages will be highlighted in regular newsletters distributed to health professionals and stakeholders (see below, Stakeholder Engagement and Consultation).
Additional educational materials which are culturally appropriate for Aboriginal and Torres Strait Islander women with a pregnancy complicated by diabetes will be developed in collaboration with the Partnership's Indigenous Reference Group and provided to healthcare services for distribution to women.
Recall and Reminder Systems Implementation staff will collaborate with health services to improve systems and enhance recall of women with a pregnancy complicated by diabetes. This will include developing and embedding appropriate postpartum and preconception care plans with reminder systems in primary care electronic health records; improvement of discharge summary templates following hospital discharge; providing postpartum summaries with guideline-based recommendations for ongoing care; and generating postpartum lists from the Diabetes in Pregnancy Clinical Register of women with a recent pregnancy complicated by hyperglycaemia for distribution to primary care services (see below, Diabetes in Pregnancy Clinical Register).

Diabetes in Pregnancy Clinical
Register Implementation of the Diabetes in Pregnancy (DIP) Clinical Register in the NT has previously been described in detail [27,43]. In brief, the register was established in the NT in 2011 and documented all consenting women residing in  the NT aged 16 years and over with any type of hyperglycaemia in pregnancy, referred by a health professional.
In the current health systems intervention, the implementation team will collaborate with health services to establish the DIP Clinical Register in FNQ, and expand its use in the NT. This will include streamlining referral processes by embedding referrals within electronic health records of health services where possible. Functions of the DIP Clinical Register include producing and distributing local postpartum reports with details of women attending specific primary healthcare services who have recently given birth, to aid with recall of women and quality improvement activities. In addition, de-identified aggregate reports including all women across the NT and FNQ are produced and distributed to healthcare providers and stakeholders as a quality improvement and epidemiological tool. Implementation staff will work with healthcare staff to use findings of aggregated DIP Clinical Register reports as quality improvement tools to inform changes in models of care and improve integration between primary and hospitalbased care.

Stakeholder Engagement and Consultation
Ongoing collaboration with stakeholders will be essential for implementation of this health systems intervention. The Partnership has established reference groups (a Clinical Reference Group and an Indigenous Reference Group in the NT, and a Partnership Working Group in FNQ), who will meet regularly throughout the health systems intervention to provide feedback on implementation activities and guide ongoing priorities of the intervention. Additional consultation will occur frequently with key health service representatives and champions. The Partnership will produce regular newsletters to provide educational messages and update health practitioners and stakeholders on project activities.

Evaluation
This health systems intervention will be evaluated using the five dimensions of the RE-AIM framework: reach (the proportion and representativeness of participants relative to target population), effectiveness (impacts of the program), adoption (uptake of the intervention), implementation (the extent to which the intervention was delivered as intended) and maintenance (the extent to  which the intervention is institutionalised into routine organisational practices and policies) [44][45][46]. Evaluation indicators for each phase of the evaluation have been developed and structured within the RE-AIM framework (Tables 3 and 4) [44]. Outcome measures include changes between baseline and post-intervention in indicators of care provided to women, including: proportion of high risk women receiving recommended glucose screening in early pregnancy; first and third trimester HbA1c in women with T2DM; and proportion of women receiving recommended postpartum care (glucose testing, breastfeeding, weight management, smoking education and discussion or prescription of contraception). We will also assess changes in health practitioner confidence in providing care, knowledge and use of relevant guidelines and referral pathways, and perception of care coordination and communication systems; and changes to health systems including referral pathways and clinical guidelines.
Changes in birth and neonatal outcomes will be assessed to determine the impact of health system changes, including: gestational age at delivery; mode of delivery; birth weight; large for gestational age; small for gestational age; macrosomia; neonatal obstetric trauma; neonatal hypoglycaemia requiring treatment; neonatal special care admission; 5-minute APGAR score less than 5; neonatal jaundice requiring treatment; and neonatal respiratory distress.
The health systems intervention will be evaluated across multiple levels (individual health practitioner, clinic, health system, community), with data collection occurring at three timepoints: 1. Baseline, prior to implementation of the intervention, to enable comparison across indicators pre-and post-intervention. 2. A qualitative interim evaluation, to be conducted at least 12 months after delivery of intervention activities has commenced across all regions, focusing on identifying and exploring barriers and enablers of implementation of and engagement with the health systems intervention. Findings from this

A mixed-methods final evaluation of outcomes and
processes of the health systems intervention, after completion of the implementation period. This evaluation will address whether the health systems intervention has improved systems of care for women during and after a pregnancy complicated by diabetes; which implementation activities have contributed to improvements; enablers and barriers impacting on the success of implementation activities; social validity of the health systems intervention (from the perspectives of healthcare providers, champions and stakeholders); and how implementation activities can be sustained beyond completion of the intervention. Quantitative data will include measures of care provided during and after pregnancy, as well as birth outcomes.

Data Collection
Data will be collected from four sources: interviews with healthcare providers and stakeholders; healthcare provider survey; cross-sectional pre-and post-intervention audits of electronic health records and the Diabetes in Pregnancy Clinical Register; and implementation operational documents. Six primary healthcare services, including one government and one community-controlled service in each of the three study regions, will be evaluation case study sites. Case study sites include both an urban and a remote service in each study region. Data will additionally be collected at the regional health service level, including regional referral hospitals and health service management.
Semi-Structured Interviews Healthcare providers and stakeholders at each of the evaluation case study sites will be interviewed during both the interim and final evaluations. Additional interviews will be conducted with relevant stakeholders at the major referral hospital within each region (Top End: Royal Darwin Hospital; Central Australia: Alice Springs Hospital; FNQ: Cairns Hospital), and with policymakers and managers at the regional health service level, as well as with members of the implementation team. These interviews will be guided by a social constructionist epistemological perspective, utilising a phenomenological approach. Interview topics will include awareness of, engagement with and utility of activities of the health systems intervention, and the impact of intervention activities on practice. Interviews with implementation staff will additionally address process measures including barriers and facilitators to implementing intervention activities, any adaptations to implementation activities and the rationale and success for these adaptations. Interviews during the final evaluation will also explore healthcare provider   Interviews confidence in providing care, perceptions of care coordination and communication and how these have changed during the health systems intervention; perceived impact of the intervention on care for women whose pregnancies are complicated by hyperglycaemia; and sustainability of implementation activities.
Survey Healthcare providers in the study regions involved in the care or care coordination of women during and/or after a pregnancy complicated by hyperglycaemia will be invited through health professional networks to complete an online survey after completion of the implementation period. Participants will be asked about their usual practice in providing care for women during and after a pregnancy complicated by diabetes; confidence and support received in providing care; satisfaction with care pathways (e.g. access to specialist and allied health care) and communication between services; and awareness and engagement with activities of the health systems intervention. Response rates will be estimated by region and profession by comparing with number of professionals per region registered with relevant professional associations, where these numbers are available. Baseline survey data for comparison was previously collected during formative work [31,37].

Audit of Electronic Health Records and Diabetes in Pregnancy Clinical
Register Primary care medical records and the DIP Clinical Register will be assessed using an independent sample, cross-sectional, pre-and post-intervention audit to determine the proportion of women receiving recommended care during and after pregnancy. In the NT, de-identified remote primary care medical records will be requested from NT Department of Health (DoH) for women with a birth due date in the final 12 months of implementation of the health systems intervention and who had a diagnosis of diabetes in pregnancy, and compared to baseline data already provided by NT DoH. In FNQ, data will be collected from the FNQ DIP Clinical Register, with baseline data including women giving birth in the first 12 months of the DIP Clinical Register's inception and post-intervention data including women giving birth in the final 12 months of the intervention. Data from primary healthcare provider electronic health records at evaluation case study sites not covered by NT DoH records (Aboriginal Community-Controlled Health Services in NT, and all FNQ evaluation sites) will also be requested for women with hyperglycaemia in pregnancy giving birth within the above time periods. Independent variables collected will include: study region; age; ethnicity; and type of diabetes. Outcome variables will include: number and timing of clinic visits (antenatal and postpartum); completion of recommended glucose screening in early pregnancy for high risk women; mean first-and thirdtrimester HbA1c; first-and third-trimester smoking status; postpartum variables within 6 months of birth including completion of glucose screening (including 75 gram OGTT, fasting glucose, HbA1c, random glucose, any glucose measure), weight, body mass index, waist circumference, breastfeeding status, smoking status and contraception use (prescribed or discussed with health practitioner); and birth and neonatal outcomes (gestational age at delivery, mode of delivery, birth weight, Implementation Operational Documents The intervention Activity Log will be reviewed, in addition to minutes of stakeholder meetings (including the Clinical and Indigenous Reference Groups) and attendance sheets and feedback forms from educational events.

Data Analysis
Qualitative Analyses Qualitative analyses of interview transcripts and operational documents will employ a hybrid inductive-deductive method. The first round of coding will be an inductive analysis; the second round of coding will involve a deductive analysis utilising the prespecified evaluation indicators. Themes will be clustered by study region, enabling comparison and contrast between regions. NVivo (version 12) will be used to assist the analysis processes.
Quantitative Analyses Survey data will be compared with baseline data collected through a prior health professional survey across the study regions, conducted during formative work for this health systems intervention [31,37]. Descriptive statistics will be reported. Changes between baseline and post-intervention for categorical variables will be analysed by Pearson's Chi-squared test or Fisher's exact test. Data on outcome variables collected from the audit of electronic healthcare records and the DIP Clinical Register will be compared between baseline and postintervention periods. Changes between pre-and postintervention in categorical variables will be analysed using Pearson's Chi-squared test; changes in continuous variables will be analysed using student's t-test. Multivariable analysis will be performed using logistic regression for categorical and linear regression for continuous variables to assess relationships with independent variables, including age, ethnicity, and diabetes type.
All quantitative analyses will be performed using STATA 15.0 (StataCorp, College Station, Texas). The threshold for statistical significance will be defined as p< 0.05 on two-tailed testing.

Discussion
Our health systems intervention is the first to our knowledge which aims to improve care both during pregnancy and the postpartum period for women across the broad spectrum of hyperglycaemia in pregnancy, including GDM, pre-existing type 2 diabetes and overt diabetes in pregnancy. As described above, published work to date has focussed largely on improving postpartum glucose screening specifically in women with GDM [22][23][24][25]; despite widespread recommendations regarding provision of multidisciplinary care during and after pregnancy for women with pre-existing diabetes, a substantial knowledge gap regarding the evaluation of models of care for women with pre-existing diabetes has been identified by others [47]. Our health systems intervention is unique as it aims to improve care for women with all forms of hyperglycaemia in pregnancy, both during and after pregnancy. In addition, this intervention focuses on healthcare providers delivering care to a population with a high burden of disease in a complex setting.
The development of this health systems intervention from formative work conducted with health professionals is a strength of the project. This intervention aims to translate the knowledge of barriers and opportunities identified in our formative findings into tangible health service improvements. Previous work of the Partnership in the NT includes improving systems of care during pregnancy and implementing the NT DIP Clinical Register [26,27]. These endeavours have resulted in the development of strong relationships between clinicians, policymakers and researchers. The current health systems intervention is strengthened through being informed by the input of these many stakeholders. Stakeholder collaboration will continue throughout implementation, further strengthening this intervention.
The mixed-methods evaluation, utilising multiple data sources, is another strength of this health systems intervention, enabling triangulation of findings. Quantitative audit data will demonstrate whether the intervention impacts on indicators of care provided to women during and after pregnancy, and on birth outcomes. Previous research has demonstrated the accuracy of routinely collected data regarding chronic conditions in electronic health records in remote Aboriginal communities, making this a useful resource for evaluation [48], together with use of our established DIP Clinical Register to assess pre-and post-intervention birth outcomes. The qualitative data will provide insights about acceptability and long-term feasibility of activities of the health systems intervention which will be essential information, both for planning sustainability of successful activities in the local setting and enabling replication in other settings.
The inclusion of multiple study regions, with different health systems operating across the regions and associated variation in organisational structure, policies and processes, is a challenge in this study, and will likely necessitate alterations in implementation activities depending on study region. However, this is also a strength of this intervention, providing an opportunity to assess how contextual factors of each region impact on the implementation and outcomes of the intervention. Information regarding contextual factors and their impact will form part of the findings of interest from this study.
The broad range of implementation activities that comprise this health systems intervention presents an additional challenge with regards to evaluating the impacts of these activities. This challenge is addressed through the use of multiple data sources within a structured evaluation framework, namely RE-AIM, which facilitates an in-depth exploration of the relative contributions of activities and their implementation to any observed outcomes.
Exploring the perspectives of Aboriginal and Torres Strait Islander women is essential in ensuring that healthcare is delivered in a way that meets these women's needs and is culturally safe and appropriate. The Partnership's Indigenous Reference Group is an important mechanism for ensuring voices of Aboriginal and Torres Strait Islander women are represented in this study. Additional planned work of the Partnership will ensure consumer perspectives are considered, namely those of Aboriginal and Torres Strait Islander women with a pregnancy complicated by hyperglycaemia, and inform future initiatives including greater engagement with consumers as participants of the intervention.
In conclusion, the Diabetes Across the Lifecourse: Northern Australia Partnership's health systems intervention has the potential to improve care for women during and after a pregnancy complicated by hyperglycaemia and intervene as early as possible in the life course to improve the health of women and their children. This study is anticipated to lead to improvements in clinician knowledge and skills in the management of hyperglycaemia in pregnancy. More broadly, this study has significance for health systems policy and implementation, particularly in populations at high risk of hyperglycaemia in pregnancy and transmission of risk to the next generation, including Indigenous peoples worldwide.