Primary Health Care Data Research Articles

Comparative study of hypertension, diabetes, dementia and smoking in military veterans and non-veterans: a quantitative study using primary healthcare data

IntroductionPrimary healthcare (PHC) patient medical records contain Systematised Nomenclature of Medicine-Clinical Terms (SNOMED-CT) that include information regarding diagnosis, demographics and veterans’ status. This study intended to identify, analyse and compare...

Cancer incidence and stage at diagnosis in children and adolescents in the Community of Madrid, 2015-2018.

Cancer is the leading cause of death in children aged 1-14 and the second in 15-19-year-old adolescents in Spain. The Paediatric Population-Based Cancer Registry of the Community of Madrid was created to monitor its incidence and survival. This study presents the incidence of childhood and adolescent cancer (0-19 years) in Madrid at a population level by sex, age group, type of tumour and stage at diagnosis. This study was a retrospective analysis of the total number of cases registered in the Paediatric Population-Based Cancer Registry of the Community of Madrid. The registry employs passive and active case finding: by cross-referencing hospital discharge data, primary healthcare data, mortality data and administrative information; and by validation of all potential incident cases through a review of electronic medical charts. All new diagnoses of malignant neoplasms, non-malignant neoplasms of the Central Nervous System, and uncertain and insitu neoplasms of the bladder, identified in 2015-2018 in individuals under age 20 residing in Madrid, were included. Patient information was collected along with tumour characteristics, including stage at diagnosis according to the Toronto Childhood Cancer Stage Guidelines. Age-specific and age-standardised incidence rates were computed with 95% confidence intervals (CI). A total of 1002 tumours were registered in 5,269,524 person-years, yielding an age-standardised rate of 192.7 (95% CI 184.3, 201.4) cases per million person-years. Male/female rate ratio was 1.1. The most common cancers across all ages comprised CNS tumours, leukaemias and lymphomas (primarily Hodgkin): 45.5 (95% CI 39.9, 51.7), 41.1 (95% CI 35.7, 47.1) and 35.8 (95% CI 30.9, 41.3) cases per million person-years, respectively. The proportion of metastatic tumours at diagnosis was similar for ages 0-14 (18.6%) and 15-19 (18.7%). This study provides a comprehensive understanding of childhood and adolescent cancer incidence in Madrid. The registry provides high-quality data and consolidates epidemiological surveillance of cancer in the region.

A "silent storm": uncovering the escalating crisis in mental healthcare for children and adolescents in Slovenia during and after the COVID-19 pandemic.

Our aim was to assess the burden of children and adolescents' mental health problems on the Slovenian outpatient healthcare system before, during and after the pandemic. In a retrospective analysis of healthcare indicators from 2008 to 2023, we analysed data from the National Institute of Public Health. Key domains included initial visits for mental and behavioural disorders (MBDs) to primary care for the population aged 0-19; the number of referrals to an initial assessment with a child and adolescent psychiatrist (CAP) at the secondary level for the population aged 0-17 along with the corresponding waiting times; and the number of urgent referrals for the population aged 0-17 to emergency mental health centres at the tertiary level. The calculations included rates per 1000 people. Descriptive statistics and diagrams were used to compare the data. Segmented linear regression analysis (SLR) was conducted on the primary healthcare data to identify the distinct temporal point indicating an increase. Comparing the average rates of the 2020-2022 period to those of the 2018-2019 period, there was a 20% increase in initial visits to primary care, a 23% increase in the referral rate to a CAP at the secondary level, and a 41% increase to the tertiary level of care. In secondary care, a four- to sevenfold increase in waiting times for the initial CAP assessment was observed between 2019 and 2023. The incidence of initial visits to primary healthcare services for MBD increased from 2008 to 2019 (average annual growth rate of 4.5%). The average annual growth rate for the 2020-2022 period tripled to 13.9%. The SLR showed that the trend of accelerating growth could have begun in 2017 for the 0-5 age group and possibly for the 15-19 age group as well. After the initial decline in 2020, all levels of the Slovenian healthcare system faced an increased burden of MBD in children and adolescents from 2021 to 2022 compared to pre-pandemic levels. Nevertheless, a portion of this increase aligns with longitudinal growing trends from 2008 onwards. Tackling the crisis requires urgent national action, significant improvement in organization, and investments in mental health.

Sex disparities in the prevalence, incidence, and management of diabetes mellitus: an Australian retrospective primary healthcare study involving 668,891 individuals

BackgroundIn Australia, diabetes is the fastest growing chronic condition, with prevalence trebling over the past three decades. Despite reported sex differences in diabetes outcomes, disparities in management and health targets remain unclear. This population-based retrospective study used MedicineInsight primary healthcare data to investigate sex differences in diabetes prevalence, incidence, management, and achievement of health targets.MethodsAdults (aged ≥ 18 years) attending 39 general practices in Western Australia were included. Diabetes incidence and prevalence were estimated by age category. Health targets assessed included body mass index (BMI), blood pressure, blood lipids, and glycated haemoglobin (HbA1c) levels. Medical management of diabetes-associated conditions was also investigated. Time-to-incident diabetes was modelled using a Weibull regression. A multilevel mixed-effects logistic regression model investigated risk-adjusted sex differences in achieving the HbA1c health target (HbA1c ≤ 7.0% (≤ 53 mmol/mol)).ResultsRecords of 668,891 individuals (53.4% women) were analysed. Diabetes prevalence ranged from 1.3% (95% confidence interval (CI) 1.2%-1.3%) in those aged < 50 years to 7.2% (95% CI 7.1%-7.3%) in those aged ≥ 50 years and was overall higher in men. In patients younger than 30 years, incidence was higher in women, with this reversing after the age of 50. Among patients with diabetes, BMI ≥ 35 kg/m2 was more prevalent in women, whereas current and past smoking were more common in men. Women were less likely than men to achieve lipid health targets and less likely to receive prescriptions for lipid, blood pressure, or glucose-lowering agents. Men with incident diabetes were 21% less likely than women to meet the HbA1c target. Similarly, ever recorded retinopathy, nephropathy, neuropathy, hypertension, dyslipidaemia, coronary heart disease, heart failure, peripheral vascular disease and peripheral artery disease were higher in men than women.ConclusionsThis research underscores variations in diabetes epidemiology and management based on sex. Tailoring diabetes management should consider the patient's sex.

Introduction of nirsevimab in Catalonia, Spain: description of the incidence of bronchiolitis and respiratory syncytial virus in the 2023/2024 season.

Respiratory syncytial virus (RSV) causes most of the cases of bronchiolitis and thousands of deaths annually, particularly in infants less than 6months old. In Catalonia (Spain), infants born between April 2023 and March 2024 aged 0-6months during their first RSV season have been candidates to receive nirsevimab, the novel monoclonal antibody against RSV, since October 2023. We aimed to analyse the dynamics of all-causes bronchiolitis diagnoses and RSV community infections in the current season and compare them to pre-nirsevimab epidemics. We collected epidemiological data from the Information System for Surveillance of Infections in Catalonia (SIVIC) on daily all-causes bronchiolitis clinical diagnoses and RSV-confirmed cases provided by rapid antigen tests in primary care practices. We calculated the rate ratio (RR) for the incidence of all-causes bronchiolitis for children aged 0-11m-old with respect to 12-35m-old between September 2014 and January 2024. We analysed the RR of the incidence of RSV-confirmed infection for 0-11m-old and 12-35m-old with respect to the > 35m-old, from January 2021 to January 2024. We then computed the relative difference of the RR, designated as percentage of reduction of risk, between season 2023/2024 and former epidemics. With a global coverage recorded rate for nirsevimab of 82.2% in January 2024, the age-specific 0-11m-old RR (95% CI) of RSV infection incidence for > 35m-old was 1.7 (1.5-2.0) in season 2023/2024. The RR (95% CI) had been 7.4 (5.6-9.9), 8.8 (6.9-11.3), and 7.1 (5.7-8.9) in 2020/2021, 2021/2022, and 2022/2023, respectively. Regarding the incidence of all-causes bronchiolitis for the 0-11m-old group compared to the 12-35m-old, the pre-pandemic (2014/2015-2019/2020) and 2022/2023 RR (95% CI) were 9.4 (9.2-9.6) and 6.0 (5.7-6.2), respectively, significantly higher than the RR of 3.6 (3.4-3.8) for the most recent season, 2023/2024. Conclusion: Concurring with the introduction of nirsevimab, the risk of RSV infection for infants aged 0-11m-old compared to > 35m-old has been reduced by 75.6% (73.4-77.5) in last season, and the risk for all-causes bronchiolitis for 12-35m-old by 61.9% (60.9-62.9) from the pre-pandemic period and by 39.8% (39.3-40.2) from the 2022/2023 epidemic, despite high RSV community transmission, especially in older infants What is Known: • RSV is responsible for approximately 70% of bronchiolitis cases and causes severe disease, particularly in infants < 6months of age. • Nirsevimab effectiveness against RSV-associated disease, particularly hospitalisations, was expected to be around 80%; other Spanish regions, such as Galicia and Valencia, and European countries including Luxembourg and Germany, have already reported good results in implementing nirsevimab to prevent RSV-associated hospitalisations and PICU stays. What is New: • We provide insight into the community incidence of RSV and all-causes bronchiolitis for season 2023/2024, when nirsevimab has been introduced to the Catalan population, using. primary healthcare data, which enabled us to assess the burden of RSV infections and bronchiolitis in the commonly seasonally saturated primary healthcare practices. • Our study reveals that the risk of all-causes bronchiolitis for infants aged 0-11m-old compared to older infants was reduced by 40% compared to the previous season and 62% compared to pre-pandemic standards, and for RSV infection it was reduced by 76%.

A Comparison of an Australian First Nations Primary Healthcare Data Specification with Potentially Preventable Hospitalisations.

Potentially Preventable Hospitalisations (PPH) is a widely used indicator of the effectiveness of non-hospital care. Specified using the International Classification of Diseases (ICD) coding, PPH comprises a suite of health conditions that could have potentially been prevented with appropriate care. The most recent edition of the National Guide to a Preventative Health Assessment for First Nations People documents the health conditions of interest to providers of primary care, many of which are not represented in PPH. Given the National Guide has been developed specifically with First Nations in mind, the aim of this research is twofold. The first aim is to formally posit the question of whether a summative measure of hospitalisations aligned diagnostically to the National Guide has value either as an alternative or complement to PPH in the context of First Nations primary health information. The second aim is to develop and present a prototype ICD-10 data specification for such a measure, referred to as the First Nations primary healthcare (FNPHC) data specification, and examine the age-standardised hospitalisation rates for FNPHC and PPH for correlations and/or differences. Age-standardised hospitalisation rates from 2016-17 to 2019-20 using both classifications were examined to assess the usefulness and relevance of summative measures of hospitalisations for informing primary care. Rates of FNPHC for principal diagnoses were between 1.5 and 2.5 times higher than those of PPH and approximately between 6 and 12 times higher for additional diagnoses. There was a strong correlation with PPH when rates were compared across all observations: jurisdictions with higher rates of PPH tended to have higher rates of hospitalisations according to the custom specification. Findings support its application as a summary measure for First Nations primary care providers. Given the policy landscape in Australia that aims to close the gap, it is imperative that measures of primary health take advantage of the concepts and application of First Nations data sovereignty and governance. The validity and cultural appropriateness of the First Nations primary health data specification needs to be further researched.

Impact of Regulatory Interventions on Ulipristal Acetate 5 mg (Esmya) Use in Spain: An Interrupted Time-Series Analysis.

Since late 2017, the use of ulipristal acetate 5 mg (UPA; Proprietary name: Esmya) has been under review in the European Union, due to an emerging hepatic risk. In February 2018 and in July 2018, the Spanish Agency of Medicines and Medical Devices and the marketing authorization holder put two risk minimization measures (RMM) in place, in order to inform about new safety information and to mitigate this risk. This study aims to assess RMM effectiveness in Spain, by performing an interrupted time-series (ITS) analyses, between 2014 and 2019. Two quasi-experimental ITS analyses to examine the use of UPA before and after the RMM release were performed: (a) an ecological study using aggregated data from a drug consumption database; and (b) a study using primary healthcare data gathered from electronic clinical records. Regulatory interventions were associated with an immediate and significant decrease level of DID (the number of DDD dispensed per 100 000 inhabitants and day) and incidence. The DID was 70% less than expected 12 months after the interventions. This value was 59% for the incidence. However, a change in the slope was not observed and the use started rising again in the last segment of the study period. Despite RMM had an immediate strong impact on UPA use, the last segment upward trend in the long-term might have been affected by the lack of comparable therapeutic alternatives. Further studies should be performed to confirm the increase trend observed and analyze subsequent measures and additional data.

No association between oral contraceptive exposure and subsequent MS: A population-based nested case–control study in primary care

Background: Conflicting data exist around oral contraceptive exposure and subsequent multiple sclerosis (MS). Objective: To use routinely collected primary healthcare data to explore the potential association between oral contraceptive exposure and subsequent MS in females at population level. Methods: We performed a nested case–control study using electronic primary care data, with complete electronic ascertainment from 1990. Logistic regression was used to evaluate associations between contraceptive exposure and MS, without and with adjusting for age, ethnicity and deprivation. Results: A total of 4455 females were included: 891 cases and 3564 controls. No association was seen between oral contraceptive exposure and subsequent MS, or between any contraceptive, combined oral contraceptive pill (COCP) or progesterone-only pill (POP) use 0–2, 2–5 or >5 years prior to MS. Conclusions: In the largest population-based study to date, we find no evidence of an association between oral contraceptive exposure and subsequent MS diagnosis.

Association of anti-diabetic drugs and COVID-19 outcomes in patients with diabetes mellitus type 2 and cardiomyopathy

There is a scarcity of information on the population with diabetes mellitus type 2 and cardiomyopathy (PDMC) in COVID-19, especially on the association between anti-diabetic medications and COVID-19 outcomes. Study is designed as a retrospective cohort analysis covering 2020 and 2021. Data from National Diabetes Registry (CroDiab) were linked to hospital data, primary healthcare data, the SARS-CoV-2 vaccination database, and the SARS-CoV-2 test results database. Study outcomes were cumulative incidence of SARS-CoV-2 positivity, COVID-19 hospitalizations, and COVID-19 deaths. For outcome predictors, logistic regression models were developed. Of 231 796 patients with diabetes mellitus type 2 in the database, 14 485 patients had cardiomyopathy. The two2-year cumulative incidence of all three studies' COVID-19 outcomes was higher in PDMC than in the general diabetes population (positivity 15.3% vs. 14.6%, p = 0.01; hospitalization 7.8% vs. 4.4%, p < 0.001; death 2.6% vs. 1.2%, p < 0.001). Sodium-Glucose Transporter 2 (SGLT-2) inhibitors therapy was found to be protective of SARS-CoV-2 infections [OR 0.722 (95% CI 0.610–0.856)] and COVID-19 hospitalizations [OR 0.555 (95% CI 0.418–0.737)], sulfonylureas to be risk factors for hospitalization [OR 1.184 (95% CI 1.029–1.362)] and insulin to be a risk factor for hospitalization [OR 1.261 (95% CI 1.046–1.520)] and death [OR 1.431 (95% CI 1.080–1.897)]. PDMC are at greater risk of acquiring SARS-CoV-2 infection and having worse outcomes than the general diabetic population. SGLT-2 inhibitors therapy was a protective factor against SARS-CoV-2 infection and against COVID-19 hospitalization, sulfonylurea was the COVID-19 hospitalization risk factor, while insulin was a risk factor for all outcomes. Further research is needed in this diabetes sub-population.

Association of anti-diabetic drugs and covid-19 outcomes in patients with diabetes mellitus type 2 and chronic kidney disease: Nationwide registry analysis.

Patients with diabetes mellitus type 2 and chronic kidney disease (T2DM-CKD) have a 5 times higher risk of developing severe SARS-CoV-2 infection than those without these 2 diseases. The goal of this study is to provide information on T2DM-CKD and COVID-19 outcomes, with an emphasis on the association with anti-diabetic medications. Study is designed as a retrospective cohort analysis covering the years 2020 and 2021. Data from the National Diabetes Registry (CroDiab) were linked to hospital data, primary healthcare data, Causes of Death Registry data, the SARS-CoV-2 vaccination database, and the SARS-CoV-2 test results database. Study outcomes were cumulative incidence of SARS-CoV-2 positivity, COVID-19 hospitalizations, and COVID-19 deaths. For outcome predictors, logistic regression models were developed. Of 231 796 patients with diabetes mellitus type 2 in the database, 7 539 were T2DM-CKD (3.25%). The 2-year cumulative incidences of all three studies' outcomes were higher in T2DM-CKD than in diabetes patients without CKD (positivity 18.1% vs. 14.4%; hospitalization 9.7% vs. 4.2%; death 3.3% vs. 1.1%, all p<0.001). For COVID-19 hospitalization, protective factors were SGLT-2 inhibitors use (OR 0.430; 95%CI 0.257-0.719) and metformin use (OR 0.769; 95% CI 0.643-0.920), risk factors were insulin use (1.411; 95%CI 1.167-1.706) and sulfonylureas use (OR 1.226; 95% CI 1.027-1.464). For SARS-CoV-2 positivity protective factors were SGLT-2 inhibitors (0.607; 95% CI 0.448-0.823), repaglinide use (OR 0.765; 95% CI 0.593-0.986) and metformin use (OR 0.857; 95% CI 0.770-0.994). DPP-4 inhibitors showed a non-significant decrease in risk for COVID-19 death (OR 0.761; 95% CI 0.568-1.019). T2DM-CKD are heavily burdened by COVID-19 disease. Our results suggest no association between antidiabetic drugs and COVID-19 death outcome while SGLT-2 and metformin show to be protective against COVID-19 hospitalization and infection, repaglinide against infection, and insulin and sulfonylureas show to be risk factors for COVID-19 hospitalization and infection. Further research in T2DM-CKD is needed.

Using machine learning techniques to predict the risk of osteoporosis based on nationwide chronic disease data

Osteoporosis is a major public health concern that significantly increases the risk of fractures. The aim of this study was to develop a Machine Learning based predictive model to screen individuals at high risk of osteoporosis based on chronic disease data, thus facilitating early detection and personalized management. A total of 10,000 complete patient records of primary healthcare data in the German Disease Analyzer database (IMS HEALTH) were included, of which 1293 diagnosed with osteoporosis and 8707 without the condition. The demographic characteristics and chronic disease data, including age, gender, lipid disorder, cancer, COPD, hypertension, heart failure, CHD, diabetes, chronic kidney disease, and stroke were collected from electronic health records. Ten different machine learning algorithms were employed to construct the predictive mode. The performance of the model was further validated and the relative importance of features in the model was analyzed. Out of the ten machine learning algorithms, the Stacker model based on Logistic Regression, AdaBoost Classifier, and Gradient Boosting Classifier demonstrated superior performance. The Stacker model demonstrated excellent performance through ten-fold cross-validation on the training set and ROC curve analysis on the test set. The confusion matrix, lift curve and calibration curves indicated that the Stacker model had optimal clinical utility. Further analysis on feature importance highlighted age, gender, lipid metabolism disorders, cancer, and COPD as the top five influential variables. In this study, a predictive model for osteoporosis based on chronic disease data was developed using machine learning. The model shows great potential in early detection and risk stratification of osteoporosis, ultimately facilitating personalized prevention and management strategies.

Association of state-level prescription drug monitoring program implementation with opioid prescribing transitions in primary care in Australia.

This study aimed to evaluate whether voluntary and mandatory prescription drug monitoring program (PDMP) use in Victoria, Australia, had an impact on prescribing behaviour, focusing on individual patients' prescribed opioid doses and transition to prescribing of nonmonitored medications. This was a retrospective cross-sectional study using routinely collected primary healthcare data. A 90-day moving average prescribed opioid dose in oral morphine equivalents was used to estimate opioid dosage. A Markov transition matrix was used to describe how patients prescribed medications transitioned between opioid dose groups and other nonopioid treatment options during 3 transition periods: transition between 2 control periods prior to PDMP implementation (T1 to T2); during the voluntary PDMP implementation (T2 to T3); and during mandatory PDMP implementation (T3 to T4). Among patients prescribed opioids in our study, we noted an increased probability of transitioning to not being prescribed opioids during the mandatory PDMP period (T3 to T4). This increase was attributed mainly to the ceasing of low-dose opioid prescribing. Membership in an opioid dose group remained relatively stable for most patients who were prescribed high opioid doses. For those who were only prescribed nonmonitored medications initially, the probability of being prescribed opioids increased during the mandatory PDMP when compared to other transition periods. The introduction of PDMP mandates appeared to have an impact on the prescribing for patients who were prescribed low-dose opioids, while its impact on individuals prescribed higher opioid doses was comparatively limited.

Participant characteristics and reasons for non-consent to health information linkage for research: experiences from the ATHENA COVID-19 study

BackgroundThe linkage of primary care, hospital and other health registry data is a global goal, and a consent-based approach is often used. Understanding the attitudes of why participants take part is important, yet little is known about reasons for non-participation. The ATHENA COVID-19 feasibility study investigated: 1) health outcomes of people diagnosed with COVID-19 in Queensland, Australia through primary care health data linkage using consent, and 2) created a cohort of patients willing to be re-contacted in future to participate in clinical trials. This report describes the characteristics of participants declining to participate and reasons for non-consent.MethodsPatients diagnosed with COVID-19 from January 1st, 2020, to December 31st, 2020, were invited to consent to having their primary healthcare data extracted from their GP into a Queensland Health database and linked to other data sets for ethically approved research. Patients were also asked to consent to future recontact for participation in clinical trials. Outcome measures were proportions of patients consenting to data extraction, permission to recontact, and reason for consent decline.ResultsNine hundred and ninety-five participants were approached and 842(85%) reached a consent decision. 581(69%), 615(73%) and 629(75%) consented to data extraction, recontact, or both, respectively. Mean (range) age of consenters and non-consenters were 50.6(22-77) and 46.1(22-77) years, respectively. Adjusting for age, gender and remoteness, older participants were more likely to consent than younger (aOR 1.02, 95%CI 1.01 to 1.03). The least socio-economically disadvantaged were more likely to consent than the most disadvantaged (aOR 2.20, 95% 1.33 to 3.64). There was no difference in consent proportions regarding gender or living in more remote regions. The main reasons for non-consent were ‘not interested in research’ (37%), ‘concerns about privacy’ (15%), ‘not registered with a GP’ (8%) and ‘too busy/no time’ (7%). ‘No reason’ was given in 20%.ConclusionYounger participants and the more socio-economically deprived are more likely to non-consent to primary care data linkage. Lack of patient interest in research, time required to participate and privacy concerns, were the most common reasons cited for non-consent. Future health care data linkage studies addressing these issues may prove helpful.

Prevalence and Characteristics of Hidradenitis Suppurativa in the Northern Finland Birth Cohort 1986 Study: A Cross-sectional Study of 2,775 Subjects.

A negative stigmatization related to hidradenitis suppurativa may prevent patients from seeking care. Thus, a large proportion of patients with hidradenitis suppurativa may be missing from studies based on hospital data. The aim of this study was to examine the prevalence and characteristics of hidradenitis suppurativa among 2,775 subjects in the Northern Finland Birth Cohort 1986 Study (NFBC1986), who were in their mid-thirties. The prevalence of hidradenitis suppurativa was 4.0% (n = 110/2,775), being higher in females (4.8%) than in males (2.5%) (p < 0.01). Of those defined as having hidradenitis suppurativa in this study, only 4 cases (n = 4/110, 3.6%) were found to have a hidradenitis suppurativa diagnosis either in the hospital (Care Register for Health Care) or in the primary healthcare data. In an adjusted model in logistic regression analyses, hidradenitis suppurativa was significantly associated with obesity (body mass index 30-55 kg/m2) (odds ratio 3.81, 95% confidence interval 2.80-5.22), female sex (1.99, 1.53-2.61) and smoking (1.56, 1.21-2.00). In addition, there was an association between hidradenitis suppurativa and self-reported poor overall health status. Hidradenitis suppurativa seems to be common at the population level and only a minority of these patients seek care for the condition.

Identifying and handling data bias within primary healthcare data using synthetic data generators

Advanced synthetic data generators can simulate data samples that closely resemble sensitive personal datasets while significantly reducing the risk of individual identification. The use of these advanced generators holds enormous potential in the medical field, as it allows for the simulation and sharing of sensitive patient data. This enables the development and rigorous validation of novel AI technologies for accurate diagnosis and efficient disease management. Despite the availability of massive ground truth datasets (such as UK-NHS databases that contain millions of patient records), the risk of biases being carried over to data generators still exists. These biases may arise from the under-representation of specific patient cohorts due to cultural sensitivities within certain communities or standardised data collection procedures. Machine learning models can exhibit bias in various forms, including the under-representation of certain groups in the data. This can lead to missing data and inaccurate correlations and distributions, which may also be reflected in synthetic data. Our paper aims to improve synthetic data generators by introducing probabilistic approaches to first detect difficult-to-predict data samples in ground truth data and then boost them when applying the generator. In addition, we explore strategies to generate synthetic data that can reduce bias and, at the same time, improve the performance of predictive models.

Measuring what counts in Aboriginal and Torres Strait Islander care: a review of general practice datasets available for assessing chronic disease care.

Background Large datasets exist in Australia that make de-identified primary healthcare data extracted from clinical information systems available for research use. This study reviews these datasets for their capacity to provide insight into chronic disease care for Aboriginal and Torres Strait Islander peoples, and the extent to which the principles of Indigenous Data Sovereignty are reflected in data collection and governance arrangements. Methods Datasets were included if they collect primary healthcare clinical information system data, collect data nationally, and capture Aboriginal and Torres Strait Islander peoples. We searched PubMed and the public Internet for data providers meeting the inclusion criteria. We developed a framework to assess data providers across domains, including representativeness, usability, data quality, adherence with Indigenous Data Sovereignty and their capacity to provide insights into chronic disease. Datasets were assessed against the framework based on email interviews and publicly available information. Results We identified seven datasets. Only two datasets reported on chronic disease, collected data nationally and captured a substantial number of Aboriginal and Torres Strait Islander patients. No dataset was identified that captured a significant number of both mainstream general practice clinics and Aboriginal Community Controlled Health Organisations. Conclusions It is critical that more accurate, comprehensive and culturally meaningful Aboriginal and Torres Strait Islander healthcare data are collected. These improvements must be guided by the principles of Indigenous Data Sovereignty and Governance. Validated and appropriate chronic disease indicators for Aboriginal and Torres Strait Islander peoples must be developed, including indicators of social and cultural determinants of health.

Cholecystectomy is associated with a higher risk of irritable bowel syndrome in the UK Biobank: a prospective cohort study.

Background: Recent studies have shown that bile acids are essential in irritable bowel syndrome (IBS) pathology, and cholecystectomy has direct effects on bile acid metabolism. However, whether cholecystectomy increases the risk of IBS remains unclear. We aimed to investigate the association between cholecystectomy and IBS risk in the UK Biobank (UKB). Methods: This study is a prospective analysis of 413,472 participants who were free of IBS, inflammatory bowel disease, cancer, or common benign digestive tract diseases. We identified incidents of IBS through self-reporting or links to primary healthcare and hospitalization data. We evaluated hazard ratios (HRs) adjusted for sociodemographic characteristics, health behaviours, comorbidities, and medications. Results: During a median follow-up period of 12.7years, we observed 15,503 new cases of IBS. Participants with a history of cholecystectomy had a 46% higher risk of IBS than those without (HR = 1.46, 95% CI: 1.32-1.60), and further subtype analysis showed that the risk of IBS with diarrhoea was significantly higher than the risk of IBS without diarrhoea (HR = 1.71, 95% CI: 1.30-2.25 vs. HR = 1.42, 95% CI: 1.28-1.58). The overall covariate-adjusted HRs for IBS were similar between the group with both cholecystectomy and gallstones (HR = 1.45, 95% CI: 1.32-1.58) and the group with cholecystectomy without gallstones (HR = 1.50, 95% CI: 1.36-1.67) when the group without both cholecystectomy and gallstones was used as a reference. The overall covariate-adjusted HR was not significantly different in the group without cholecystectomy with gallstones (HR = 1.18, 95% CI: 0.95-1.47). The positive association of cholecystectomy with IBS risk did not change when stratifying the data based on age, sex, BMI, smoking, alcohol consumption, healthy diet, quality sleep, physical activity, type 2 diabetes, hypertension, hyperlipidaemia, mental illness, NSAID intake, or acid inhibitor intake. Sensitivity analyses, including propensity score matching analysis and lagging the exposure for two or fouryears, indicated that the effects were robust. Conclusion: Cholecystectomy was associated with a higher risk of IBS, especially IBS with diarrhoea. Additional prospective randomized controlled and experimental studies are warranted to further validate the association and to explore the relevant biological mechanisms.

Early detection of respiratory disease outbreaks through primary healthcare data

Early detection of respiratory disease outbreaks through primary healthcare data

An ontology-based approach to designing a NoSQL database for semi-structured and unstructured health data.

With the advent of ICT-based healthcare applications, various formats of health data are generated every day in huge volume. Such data, consisting of unstructured, semi-structured and structured data, has every characteristic of Big data. NoSQL databases are generally preferred for storing such type of health data with the objective of improving query performance. However, for efficient retrieval and processing of Big Health Data and for resource optimization, suitable data models and design of the NoSQL databases are important requirements. Unlike relational databases, no standard methods or tools exist for NoSQL database design. In this work, we adopt an ontology-based schema design approach. We propose that an ontology, which captures the domain knowledge, be used for developing a health data model. An ontology for primary healthcare is described in this paper. We also propose an algorithm for designing the schema of a NoSQL database, keeping in mind the characteristics of the target NoSQL store, using a related ontology, a sample query set, some statistical information of the queries, and performance requirements of the query set. The ontology proposed by us for primary healthcare domain and the above mentioned algorithm along with a set of queries are used for generating a schema targeting MongoDB datastore. The performance of the proposed design is compared with a relational model developed for the same primary healthcare data and the effectiveness of our proposed approach is demonstrated. The entire experiment has been carried out on MongoDB cloud platform.

Emergency presentation prior to lung cancer diagnosis: A national-level examination of disparities and survival outcomes

ObjectivesA recent multinational investigation of emergency presentation within 30 days of cancer diagnosis, conducted within the International Cancer Benchmarking Programme (ICBP), observed that New Zealand had the highest rate of emergency presentation prior to lung cancer diagnosis compared to other similar countries. Here we use national-level health data to further investigate these trends, focussing on ethnic disparities in emergency presentation prior to lung cancer diagnosis. We have also compared survival outcomes between those who had an emergency presentation in the preceding 30 days to those who did not. Materials and methodsOur study included all lung cancer registrations between 2007 and 2019 on the New Zealand Cancer Registry (N = 27,869), linked to national hospitalisation and primary healthcare data. We used descriptive (crude and age-standardised proportions) and logistic regression (crude and adjusted odds ratios) analyses to examine primary care access prior to cancer diagnosis, emergency hospitalisation up to and including 30 days prior to diagnosis, and one-year mortality post-diagnosis, both for the total population and between ethnicities. Regression models adjusted for age, sex, deprivation, rurality, comorbidity, tumour type and stage. ResultsWe found stark disparities by ethnic group, with 62% of Pacific peoples and 54% of Māori having an emergency presentation within 30 days prior to diagnosis, compared to 47% of Europeans. These disparities remained after adjusting for multiple covariates including comorbidity and deprivation (adj. OR: Māori 1.21, 95% CI 1.13–1.30; Pacific 1.50, 95% CI 1.31–1.71). Emergency presentation was associated with substantially poorer survival outcomes across ethnic groups (e.g. 1-year mortality for Māori: no emergency presentation 50%, emergency presentation 79%; adj. OR 2.40, 95% CI 2.10–2.74). ConclusionsThese observations reinforce the need for improvements in the early detection of lung cancer, particularly for Māori and Pacific populations, with a view to preventing diagnosis of these cancers in an emergency setting.