Data Resource Profile: Paediatric Regional Healthcare Information Platform in Halland, Sweden

Deceased The Paediatric Regional Healthcare Information Platform (p-RHIP) was established in Halland, Sweden, to enhance paediatric health by identifying early risk factors in children aged 0–6 years. It integrates comprehensive, routinely collected data from multiple healthcare sources, including three hospitals, two emergency departments, 48 primary care units, and national registers, allowing for in-depth, real-world evidence research that supports personalized care and early interventions. At baseline, it includes data from 61 040 children, collected from 2009 to 2022 and is updated weekly. It is visit- and patient-centred, organizing data around healthcare encounters and linking this with demographic, socioeconomic, and cost information. The data are owned by Region Halland (RH), housed within RH’s firewalls and supported by strong in-house computing resources, enabling advanced analyses without the need to export data. Access for operational analyses is approved by administrators for each task according to a pre-specified process. Interest for research collaborations can be addressed to the corresponding author ([email protected]). The landscape of modern healthcare is increasingly shaped by evidence-based medicine, where informed decisions and clinical practices rely on rigorous scientific research and data analysis [1]. High-quality healthcare data are fundamental to evidence-based medicine, providing the empirical foundation for medical professionals to understand health outcomes and treatment efficacy. This reliance on data has sparked a growing trend of using electronic health records (EHRs) beyond their primary purpose for analytical insights [2]. However, leveraging healthcare data for research purposes presents formidable challenges. First, there is the technical hurdle of interoperability. Healthcare systems often operate on disparate platforms, making it difficult to seamlessly share and analyse data across different organizations. For example, a patient’s medical history recorded in a primary care clinic may not readily integrate with data from a specialist hospital. Second, there is the ethical and legal imperative to safeguard patient privacy and ensure responsible data usage. Healthcare data contain sensitive information, and their secondary use for research must adhere to strict regulations and ethical guidelines to protect patient confidentiality. Overcoming these challenges, we developed the Regional Healthcare Information Platform (RHIP) [3], a comprehensive data system that collects population-wide healthcare data in the province of Halland. Halland, located in the southwest of Sweden, is the seventh largest province (of 21) by population. This article introduces paediatric RHIP (p-RHIP), a specialized extension of RHIP for children aged 0–6 years, established in 2024 with weekly data updates and retrospective since 2009. While p-RHIP focuses on data from ages 0 to 6, data beyond age 6 continue to be linked within the overall RHIP infrastructure. By integrating comprehensive obstetrics, growth data, and other RHIP data, p-RHIP serves as a central hub for internal quality control, process improvement, decision-making, and high-quality, data-driven paediatric research. At its core, p-RHIP aims to identify early risk factors in children’s well-being, facilitating personalized care and timely interventions [4]. By refining care guidelines and enabling early interventions, the platform seeks to enhance children’s outcomes—a vital pursuit for the future health of society. The p-RHIP is populated with routinely collected EHR data from three hospitals, two emergency departments, and 46 primary care units (including private healthcare providers) in the Halland province, serving around 344 000 inhabitants as of December 2023 [5]. These primary care units include integrated general practices and child health services (CHS). In Sweden, CHS provide free, nationwide healthcare for children under 6. Almost all children in Sweden attend regular visits at CHS, led by registered specialist nurses in primary care [6]. These visits, typically numbering 15–16 in the first year, include check-ups for growth monitoring, psychomotor and language development, vision and hearing screenings, vaccinations, and lifestyle advice. National registers also complement these data, offering a comprehensive view of paediatric health in the region [7]. Growth variables such as weight, length/height, and head circumference, routinely collected in child healthcare centres, serve as indicators of overall health and can reveal underlying diseases and psychosocial issues. Additionally, pre-birth and birth data, linked with parental information, enrich the person-level dataset, facilitating analysis of how prenatal factors influence child growth, along with socioeconomic data. Figure 1 provides a high-level layout of the platform. It is a visit- and patient-centred platform, organizing seven groups of data tables: The paediatric regional healthcare information platform in Halland, Sweden. Platform IDs are unique for every child and, along with Visit IDs and date-time data, are used to link different tables within the platform. Org ID refers to Organization IDs within the system. *Platform IDs can be internally converted back to Personal IDs, for the extraction of additional data from national registers. Visit tables define unique hospital and clinic encounters, including primary care, outpatient specialist, psychiatry care, emergency, inpatient, and ambulance visits. These encounters encompass both physical and virtual visits, as well as group and home visits. Detailed visit tables at the patient level elaborate on diagnoses, procedures, notes, medications, vitals, complaints, etc., for each visit. Detailed visit tables at the organization level provide information on where and with whom (pseudoanonymized healthcare staff ID) the encounter took place. Resource tables are linked via Organization IDs detailing occupancy and capacity status at department levels. Detailed patient tables contain demographic information such as sex, age, socioeconomic status of parents/guardians, primary care units, and corresponding geography IDs. Obstetrics data from pre- and perinatal times supplement the dataset by linking Platform ID with the corresponding mother’s ID. Cost tables describe costs allocated at the patient encounter level across all care venues and take into account resources and utilized capacity. p-RHIP serves as a centralized repository that is updated weekly with a vast array of healthcare data sourced from the Region’s Data Warehouse. The Data Warehouse gathers information from over 40 different IT systems spanning various care organizations. This aggregation is made possible through the unique personal ID assigned to each registered individual in Sweden, which facilitates the seamless linkage of records from disparate sources. Notably, VAS (Vårdadministrativt system) [8], which includes usual clinical data like diagnoses, concerns, and clinical notes, and Obstetrix-Milou [9], which primarily contributes midwifery and child-birth data from maternity wards, is a major contributor to p-RHIP and is extensively utilized by healthcare providers across various facilities for clinical data entry. p-RHIP is structured as a SQL database housed within the Region’s firewalls. To safeguard confidentiality and privacy, stringent measures are implemented. Patient information undergoes pseudo-anonymization, removing identifiable features such as names, addresses, and telephone numbers before being loaded into p-RHIP. Swedish Personal IDs are also replaced with platform IDs, with the key to map these IDs securely managed at the regional level. This key is accessible only for specific integration needs, such as retrieving additional data from national registers. Data are treated according to the routines of Region Halland (RH) as a healthcare provider. The platform has been reported to national authorities according to Swedish regulations. The content is owned by RH and maintained on its premises. Access to identifiable data is strictly controlled and limited to authorized IT personnel at RH. Researchers can only access data through internal client systems protected by regional IT firewalls, following a well-defined and tested process to ensure compliance with Swedish and EU data protection and privacy laws [10]. Access is granted exclusively for approved projects (e.g. EKAT 0–6, detailed below) and is restricted to the specific population definition and variables authorized for each project. This is usually achieved by creating an SQL view of p-RHIP with the approved population data, essentially a subset of p-RHIP. All analyses are performed on-site, where RH provides secure and high-end computing resources. To demonstrate key statistics about the p-RHIP platform, from here on, we present data on 61 040 children born between 2009 and 2022 and registered in Halland between the ages of 0 and 6 years, in accordance with ethics approval (2022-06743-01). Figure 2 illustrates the distribution of birth cohorts, separated by sex. Table 1 presents the coverage of key variables deemed critical for paediatric analyses in p-RHIP, along with event counts that highlight the extensive scale of the platform. Birth cohorts. Coverage of key variables and recorded event counts in p-RHIP Coverage of key variables and recorded event counts in p-RHIP p-RHIP builds on RHIP’s success in adult research and enables paediatric studies by a multidisciplinary team, leveraging real-world evidence. The data support diverse research directions. The database holds extensive children’s clinical data, including symptoms, medical history, infant nutrition, and socioeconomic factors, which researchers can use to develop advanced algorithms and screening tools for early health warning signs, such as disease onset. Childhood obesity is a global health issue with medical, social, and economic challenges for children, parents, and communities [11, 12]. Prevention and early intervention are crucial in childhood growth monitoring, where p-RHIP offers potential improvements. Table 2 presents diagnoses in children at different ages. Analysing these data may reveal trends and risk factors, enabling proactive healthcare strategies and personalized interventions for better health outcomes. Distribution of International Classification of Diseases, 10th Edition (ICD-10) codes in p-RHIP between January 2009 and December 2022 for the 61 040 children in the paediatric cohort Distribution of International Classification of Diseases, 10th Edition (ICD-10) codes in p-RHIP between January 2009 and December 2022 for the 61 040 children in the paediatric cohort Unlike typical EHR data, which is recorded only when a patient seeks care [13], CHS collect routine data, enabling the identification of anomalies by defining normal growth patterns. Supplementary Figure S1 illustrates length/height and weight distributions in p-RHIP over 3-month intervals for the 61 040 children. These data support evidence-based growth research and enable personalized growth trajectories by integrating factors like parental heights, birth details, and socioeconomic status. These individualized standards can help healthcare professionals monitor growth more accurately, potentially detecting early deviations that may serve as biomarkers for future diagnoses. Similarly, researchers may use the database to conduct healthcare economic analyses, examining the financial aspects of healthcare systems and services. This includes data on healthcare expenditures, resource utilization, and patient outcomes. p-RHIP utilizes two cost calculation methods: cost per patient (CPP) [14] and patient encounter cost (PEC) [15]. The principal difference between the two is that the CPP model distributes costs that cannot be directly allocated to a specific visit or specific service to visits by dividing these costs by the actual number of care contacts. The PEC model assigns these costs to visits by dividing them with the maximum number of contacts that can be performed at full capacity. By integrating these datasets, researchers may conduct cost-effectiveness studies, modelling the economic impact of different healthcare interventions and policies. Additionally, longitudinal analyses may track trends in healthcare spending over time and identify areas for cost containment or efficiency improvements. Furthermore, economic analyses can inform decision-makers about resource allocation, helping to prioritize investments in healthcare programs and initiatives that provide the most value for money. Socioeconomic data linked to demographic statistical areas (DeSO) in Sweden’s p-RHIP framework are crucial for healthcare and epidemiological analysis. DeSO divides Sweden into 5984 neighbourhoods, each housing 653–4243 residents, to track segregation and socioeconomic conditions [16]. These data highlight disparities in healthcare access and outcomes, aiding in early interventions to mitigate health risks tied to income and education. Figure 3, adapted from [17], shows the distribution of children in p-RHIP across economic quintiles (Q1–Q5), reflecting national socioeconomic trends. Children are more likely to live in low-income households (33% of children aged 0–6 years versus 25% of the total population), shifting distributions from Q1 to Q5 [18]. In Halland, a socioeconomically higher-status region compared with the national distribution according to Statistics Sweden [18], 28% of children are in low-income households, creating a reverse shift from Q5 to Q1. Distribution of children by economic quintiles and sex. An example of an ongoing research project on p-RHIP is EKAT, which stands for Evidensbaserad Kunskap om Avvikande Tillväxt hos barn 0–6 år in Swedish, or Evidence-based knowledge about deviant growth in children 0–6 years. This project, approved by the Swedish Ethical Review Authority (2022-06743-01), focuses on children in p-RHIP who were born between 2009 and 2022 and registered in Halland until the age of 6. The aim is to develop evidence-based guidelines for investigating growth deviations in young children. The project identifies and compares growth deviations using various criteria, including current expert recommendations, Finnish algorithms [19], Quadratic-Exponential-Pubertal-Stop (QEPS) growth models [20–23], and machine learning algorithms. It also examines the relationship between growth patterns and ongoing and future illnesses and psychosocial problems to develop predictive models for individual health risks. The long-term goal is to integrate the results into medical record systems and apps, improving children’s health and medical care. Additionally, the project investigates how sex and socioeconomic factors affect growth, referrals, assessments, treatments, and healthcare consumption, aiming to develop cost-effective, individualized care programs for prevention. The key strengths of the platform lie in the highly granular coverage of individual children and provider level data (both laterally, across multiple care units and longitudinally, over long periods of time) and deep integration of socioeconomic and cost data. In addition, with appropriate approvals already in place, data are linked to external national high-quality registries (e.g. containing causes of deaths and drugs picked up at pharmacies) that are abundant in Sweden. Thus, p-RHIP empowers researchers to focus on healthcare value—defined as the health outcomes attained per Euro spent [24]. By harnessing its wealth of information, researchers and healthcare professionals can identify patterns, trends, and interventions that yield the highest health outcomes relative to the resources invested. The platform also integrates international coding schemas such as the International Classification of Diseases, 10th Revision (ICD-10), for diagnoses, procedures, and Diagnosis-Related Group codes, along with the Anatomical Therapeutic Chemical classification for medications. This enhances the platform’s compatibility and interoperability with global healthcare standards, facilitating comparative analyses across different healthcare systems and regions. A standout feature of the platform is its commitment to privacy, confidentiality, and data security, ensuring that data remain on-site without the need for exporting to third parties. It also offers abundant high-performance computing resources, including an NVIDIA DGX platform, enabling advanced artificial intelligence research and sophisticated analyses on large datasets. Despite its comprehensive nature, p-RHIP relies on routinely collected data, making it susceptible to inherent biases akin to those encountered in observational studies. These biases can present challenges in establishing causal relationships in clinical research. Another limitation arises from the geographical scope of p-RHIP, which currently covers Halland. This may affect data completeness for children residing in border areas, particularly north of Halland, who may receive primary care in the Västra Götaland region despite being registered in Halland. Furthermore, p-RHIP only includes growth data for the first 6 years of life, collected by the CHS. Growth data for children older than 6 primarily stem from school healthcare administered by municipalities, which are not currently integrated with p-RHIP. To address this limitation, future efforts aim to incorporate school and municipality data into the platform. Additionally, p-RHIP currently lacks full integration with Picture Archiving and Communication Systems, including only the headers of radiology examinations performed. Moreover, the platform is primarily derived from healthcare systems, limiting the inclusion of social variables at the individual level. These points highlight current limitations of p-RHIP and underscore areas for future enhancement and integration efforts. Access for operational analyses is approved by administrators for each task according to a pre-specified process. Gaining research access to the platform requires an approved project and analytical plan from the Swedish Ethical Review Authority in Sweden. Researchers are required to sign a confidentiality and secrecy agreement with RH before gaining access. For studies that require data from external sources, data are available with additional approvals. The agreements have been carefully formulated after iterative legal reviews in accordance with the General Data Protection Regulation (2916/679) and other applicable laws. Interest in research collaborations can be addressed to the corresponding author ([email protected]). All statistical data reported in this paper comply with the ethical approval (2022-06743-01) granted by the Swedish Ethical Review Authority. We are deeply grateful to Professor Kerstin Albertsson-Wikland, a remarkable pioneer in paediatric growth research, who died on 8 October 2024, engaged in research until her final day. She made contributions to the initiation, conceptualization and feedback on the writing of this study. We are very grateful to Associate Professor and Neonatologist Aimon Niklasson (Gothenburg University and former senior consultant at the Neonatal department, Queens Silvias hospital, Gothenburg) for valuable thoughts regarding early growth and perinatal variables. We thank Midwife Annika Neij at the Obstetrics department, Halland Hospital Halmstad, for her invaluable assistance in extracting pre-birth and birth data. We are grateful to the head of the Department of Research and Development, Region Halland, Dr Magnus Clarin, for guaranteeing the opportunity for this research. We are to Dr regional at the department at the Västra Götaland for facilitating the of between University and Region of Halland. Additionally, we the of Halland, who their data to RH’s research platform. has a crucial in research in this and the research performed the statistics and the All to the of the the All and the Supplementary data is available at of The data platform is a supported by Region Halland, Sweden. Research projects are from and private in Sweden for collaborations can be addressed to the corresponding author ([email protected]). was used to the and of this