Association between covid-19 vaccination, SARS-CoV-2 infection, and risk of immune mediated neurological events: population based cohort and self-controlled case series analysis

OBJECTIVE

To study the association between covid-19 vaccines, SARS-CoV-2 infection, and risk of immune mediated neurological events.

DESIGN

Population based historical rate comparison study and self-controlled case series analysis.

SETTING

Primary care records from the United Kingdom, and primary care records from Spain linked to hospital data.

PARTICIPANTS

8 330 497 people who received at least one dose of covid-19 vaccines ChAdOx1 nCoV-19, BNT162b2, mRNA-1273, or Ad.26.COV2.S between the rollout of the vaccination campaigns and end of data availability (UK: 9 May 2021; Spain: 30 June 2021). The study sample also comprised a cohort of 735 870 unvaccinated individuals with a first positive reverse transcription polymerase chain reaction test result for SARS-CoV-2 from 1 September 2020, and 14 330 080 participants from the general population.

MAIN OUTCOME MEASURES

Outcomes were incidence of Bell's palsy, encephalomyelitis, Guillain-Barré syndrome, and transverse myelitis. Incidence rates were estimated in the 21 days after the first vaccine dose, 90 days after a positive test result for SARS-CoV-2, and between 2017 and 2019 for background rates in the general population cohort. Indirectly standardised incidence ratios were estimated. Adjusted incidence rate ratios were estimated from the self-controlled case series.

RESULTS

The study included 4 376 535 people who received ChAdOx1 nCoV-19, 3 588 318 who received BNT162b2, 244 913 who received mRNA-1273, and 120 731 who received Ad26.CoV.2; 735 870 people with SARS-CoV-2 infection; and 14 330 080 people from the general population. Overall, post-vaccine rates were consistent with expected (background) rates for Bell's palsy, encephalomyelitis, and Guillain-Barré syndrome. Self-controlled case series was conducted only for Bell's palsy, given limited statistical power, but with no safety signal seen for those vaccinated. Rates were, however, higher than expected after SARS-CoV-2 infection. For example, in the data from the UK, the standardised incidence ratio for Bell's palsy was 1.33 (1.02 to 1.74), for encephalomyelitis was 6.89 (3.82 to 12.44), and for Guillain-Barré syndrome was 3.53 (1.83 to 6.77). Transverse myelitis was rare (<5 events in all vaccinated cohorts) and could not be analysed.

CONCLUSIONS

No safety signal was observed between covid-19 vaccines and the immune mediated neurological events of Bell's palsy, encephalomyelitis, Guillain-Barré syndrome, and transverse myelitis. An increased risk of Bell's palsy, encephalomyelitis, and Guillain-Barré syndrome was, however, observed for people with SARS-CoV-2 infection.

Leveraging electronic health records for data science: common pitfalls and how to avoid them

Analysis of electronic health records (EHRs) is an increasingly common approach for studying real-world patient data. Use of routinely collected data offers several advantages compared with other study designs, including reduced administrative costs, the ability to update analysis as practice patterns evolve, and larger sample sizes. Methodologically, EHR analysis is subject to distinct challenges because data are not collected for research purposes. In this Viewpoint, we elaborate on the importance of in-depth knowledge of clinical workflows and describe six potential pitfalls to be avoided when working with EHR data, drawing on examples from the literature and our experience. We propose solutions for prevention or mitigation of factors associated with each of these six pitfalls-sample selection bias, imprecise variable definitions, limitations to deployment, variable measurement frequency, subjective treatment allocation, and model overfitting. Ultimately, we hope that this Viewpoint will guide researchers to further improve the methodological robustness of EHR analysis.

Comparison of Machine Learning Methods and Conventional Logistic Regressions for Predicting Gestational Diabetes Using Routine Clinical Data: A Retrospective Cohort Study

BACKGROUND

Gestational diabetes mellitus (GDM) contributes to adverse pregnancy and birth outcomes. In recent decades, extensive research has been devoted to the early prediction of GDM by various methods. Machine learning methods are flexible prediction algorithms with potential advantages over conventional regression.

OBJECTIVE

The purpose of this study was to use machine learning methods to predict GDM and compare their performance with that of logistic regressions.

METHODS

We performed a retrospective, observational study including women who attended their routine first hospital visits during early pregnancy and had Down's syndrome screening at 16-20 gestational weeks in a tertiary maternity hospital in China from 2013.1.1 to 2017.12.31. A total of 22,242 singleton pregnancies were included, and 3182 (14.31%) women developed GDM. Candidate predictors included maternal demographic characteristics and medical history (maternal factors) and laboratory values at early pregnancy. The models were derived from the first 70% of the data and then validated with the next 30%. Variables were trained in different machine learning models and traditional logistic regression models. Eight common machine learning methods (GDBT, AdaBoost, LGB, Logistic, Vote, XGB, Decision Tree, and Random Forest) and two common regressions (stepwise logistic regression and logistic regression with RCS) were implemented to predict the occurrence of GDM. Models were compared on discrimination and calibration metrics.

RESULTS

In the validation dataset, the machine learning and logistic regression models performed moderately (AUC 0.59-0.74). Overall, the GBDT model performed best (AUC 0.74, 95% CI 0.71-0.76) among the machine learning methods, with negligible differences between them. Fasting blood glucose, HbA1c, triglycerides, and BMI strongly contributed to GDM. A cutoff point for the predictive value at 0.3 in the GBDT model had a negative predictive value of 74.1% (95% CI 69.5%-78.2%) and a sensitivity of 90% (95% CI 88.0%-91.7%), and the cutoff point at 0.7 had a positive predictive value of 93.2% (95% CI 88.2%-96.1%) and a specificity of 99% (95% CI 98.2%-99.4%).

CONCLUSION

In this study, we found that several machine learning methods did not outperform logistic regression in predicting GDM. We developed a model with cutoff points for risk stratification of GDM.

Risk of bias in observational studies using routinely collected data of comparative effectiveness research: a meta-research study

BACKGROUND

To assess the completeness of reporting, research transparency practices, and risk of selection and immortal bias in observational studies using routinely collected data for comparative effectiveness research.

METHOD

We performed a meta-research study by searching PubMed for comparative effectiveness observational studies evaluating therapeutic interventions using routinely collected data published in high impact factor journals from 01/06/2018 to 30/06/2020. We assessed the reporting of the study design (i.e., eligibility, treatment assignment, and the start of follow-up). The risk of selection bias and immortal time bias was determined by assessing if the time of eligibility, the treatment assignment, and the start of follow-up were synchronized to mimic the randomization following the target trial emulation framework.

RESULT

Seventy-seven articles were identified. Most studies evaluated pharmacological treatments (69%) with a median sample size of 24,000 individuals. In total, 20% of articles inadequately reported essential information of the study design. One-third of the articles (n = 25, 33%) raised some concerns because of unclear reporting (n = 6, 8%) or were at high risk of selection bias and/or immortal time bias (n = 19, 25%). Only five articles (25%) described a solution to mitigate these biases. Six articles (31%) discussed these biases in the limitations section.

CONCLUSION

Reporting of essential information of study design in observational studies remained suboptimal. Selection bias and immortal time bias were common methodological issues that researchers and physicians should be aware of when interpreting the results of observational studies using routinely collected data.

Epidemiology of pre-existing multimorbidity in pregnant women in the UK in 2018: a population-based cross-sectional study

BACKGROUND

Although maternal death is rare in the United Kingdom, 90% of these women had multiple health/social problems. This study aims to estimate the prevalence of pre-existing multimorbidity (two or more long-term physical or mental health conditions) in pregnant women in the United Kingdom (England, Northern Ireland, Wales and Scotland).

STUDY DESIGN

Pregnant women aged 15-49 years with a conception date 1/1/2018 to 31/12/2018 were included in this population-based cross-sectional study, using routine healthcare datasets from primary care: Clinical Practice Research Datalink (CPRD, United Kingdom, n = 37,641) and Secure Anonymized Information Linkage databank (SAIL, Wales, n = 27,782), and secondary care: Scottish Morbidity Records with linked community prescribing data (SMR, Tayside and Fife, n = 6099). Pre-existing multimorbidity preconception was defined from 79 long-term health conditions prioritised through a workshop with patient representatives and clinicians.

RESULTS

The prevalence of multimorbidity was 44.2% (95% CI 43.7-44.7%), 46.2% (45.6-46.8%) and 19.8% (18.8-20.8%) in CPRD, SAIL and SMR respectively. When limited to health conditions that were active in the year before pregnancy, the prevalence of multimorbidity was still high (24.2% [23.8-24.6%], 23.5% [23.0-24.0%] and 17.0% [16.0 to 17.9%] in the respective datasets). Mental health conditions were highly prevalent and involved 70% of multimorbidity CPRD: multimorbidity with ≥one mental health condition/s 31.3% [30.8-31.8%]). After adjusting for age, ethnicity, gravidity, index of multiple deprivation, body mass index and smoking, logistic regression showed that pregnant women with multimorbidity were more likely to be older (CPRD England, adjusted OR 1.81 [95% CI 1.04-3.17] 45-49 years vs 15-19 years), multigravid (1.68 [1.50-1.89] gravidity ≥ five vs one), have raised body mass index (1.59 [1.44-1.76], body mass index 30+ vs body mass index 18.5-24.9) and smoked preconception (1.61 [1.46-1.77) vs non-smoker).

CONCLUSION

Multimorbidity is prevalent in pregnant women in the United Kingdom, they are more likely to be older, multigravid, have raised body mass index and smoked preconception. Secondary care and community prescribing dataset may only capture the severe spectrum of health conditions. Research is needed urgently to quantify the consequences of maternal multimorbidity for both mothers and children.

Treatment effects in randomised trials using routinely collected data for outcome assessment versus traditional trials: meta-research study

OBJECTIVE

To compare effect estimates of randomised clinical trials that use routinely collected data (RCD-RCT) for outcome ascertainment with traditional trials not using routinely collected data.

DESIGN

Meta-research study.

DATA SOURCE

Studies included in the same meta-analysis in a Cochrane review.

ELIGIBILITY CRITERIA FOR STUDY SELECTION

Randomised clinical trials using any type of routinely collected data for outcome ascertainment, including from registries, electronic health records, and administrative databases, that were included in a meta-analysis of a Cochrane review on any clinical question and any health outcome together with traditional trials not using routinely collected data for outcome measurement.

REVIEW METHODS

Effect estimates from trials using or not using routinely collected data were summarised in random effects meta-analyses. Agreement of (summary) treatment effect estimates from trials using routinely collected data and those not using such data was expressed as the ratio of odds ratios. Subgroup analyses explored effects in trials based on different types of routinely collected data. Two investigators independently assessed the quality of each data source.

RESULTS

84 RCD-RCTs and 463 traditional trials on 22 clinical questions were included. Trials using routinely collected data for outcome ascertainment showed 20% less favourable treatment effect estimates than traditional trials (ratio of odds ratios 0.80, 95% confidence interval 0.70 to 0.91, I2=14%). Results were similar across various types of outcomes (mortality outcomes: 0.92, 0.74 to 1.15, I2=12%; non-mortality outcomes: 0.71, 0.60 to 0.84, I2=8%), data sources (electronic health records: 0.81, 0.59 to 1.11, I2=28%; registries: 0.86, 0.75 to 0.99, I2=20%; administrative data: 0.84, 0.72 to 0.99, I2=0%), and data quality (high data quality: 0.82, 0.72 to 0.93, I2=0%).

CONCLUSIONS

Randomised clinical trials using routinely collected data for outcome ascertainment show smaller treatment benefits than traditional trials not using routinely collected data. These differences could have implications for healthcare decision making and the application of real world evidence.

Prediction of obstetrical and fetal complications using automated electronic health record data

An increasing number of delivering women experience major morbidity and mortality. Limited work has been done on automated predictive models that could be used for prevention. Using only routinely collected obstetrical data, this study aimed to develop a predictive model suitable for real-time use with an electronic medical record. We used a retrospective cohort study design with split validation. The denominator consisted of women admitted to a delivery service. The numerator consisted of women who experienced a composite outcome that included both maternal (eg, uterine rupture, postpartum hemorrhage), fetal (eg, stillbirth), and neonatal (eg, hypoxic ischemic encephalopathy) adverse events. We employed machine learning methods, assessing model performance using the area under the receiver operator characteristic curve and number needed to evaluate. A total of 303,678 deliveries took place at 15 study hospitals between January 1, 2010, and March 31, 2018, and 4130 (1.36%) had ≥1 obstetrical complication. We employed data from 209,611 randomly selected deliveries (January 1, 2010, to March 31, 2017) as a derivation dataset and validated our findings on data from 52,398 randomly selected deliveries during the same time period (validation 1 dataset). We then applied our model to data from 41,669 deliveries from the last year of the study (April 1, 2017, to March 31, 2018 [validation 2 dataset]). Our model included 35 variables (eg, demographics, vital signs, laboratory tests, progress of labor indicators). In the validation 2 dataset, a gradient boosted model (area under the receiver operating characteristic curve or c statistic, 0.786) was slightly superior to a logistic regression model (c statistic, 0.778). Using an alert threshold of 4.1%, our final model would flag 16.7% of women and detect 52% of adverse outcomes, with a number needed to evaluate of 20.9 and 0.455 first alerts per day per 1000 annual deliveries. In conclusion, electronic medical record data can be used to predict obstetrical complications. The clinical utility of these automated models has not yet been demonstrated. To conduct interventions to assess whether using these models results in patient benefit, future work will need to focus on the development of clinical protocols suitable for use in interventions.

Accessing routinely collected health data to improve clinical trials: recent experience of access

BACKGROUND

Routinely collected electronic health records (EHRs) have the potential to enhance randomised controlled trials (RCTs) by facilitating recruitment and follow-up. Despite this, current EHR use is minimal in UK RCTs, in part due to ongoing concerns about the utility (reliability, completeness, accuracy) and accessibility of the data. The aim of this manuscript is to document the process, timelines and challenges of the application process to help improve the service both for the applicants and data holders.

METHODS

This is a qualitative paper providing a descriptive narrative from one UK clinical trials unit (MRC CTU at UCL) on the experience of two trial teams' application process to access data from three large English national datasets: National Cancer Registration and Analysis Service (NCRAS), National Institute for Cardiovascular Outcomes Research (NICOR) and NHS Digital to establish themes for discussion. The underpinning reason for applying for the data was to compare EHRs with data collected through case report forms in two RCTs, Add-Aspirin (ISRCTN 74358648) and PATCH (ISRCTN 70406718).

RESULTS

The Add-Aspirin trial, which had a pre-planned embedded sub-study to assess EHR, received data from NCRAS 13 months after the first application. In the PATCH trial, the decision to request data was made whilst the trial was recruiting. The study received data after 8 months from NICOR and 15 months for NHS Digital following final application submission. This concluded in May 2020. Prior to application submission, significant time and effort was needed particularly in relation to the PATCH trial where negotiations over consent and data linkage took many years.

CONCLUSIONS

Our experience demonstrates that data access can be a prolonged and complex process. This is compounded if multiple data sources are required for the same project. This needs to be factored in when planning to use EHR within RCTs and is best considered prior to conception of the trial. Data holders and researchers are endeavouring to simplify and streamline the application process so that the potential of EHR can be realised for clinical trials.

Access to routinely collected health data for clinical trials - review of successful data requests to UK registries

BACKGROUND

Clinical trials generally each collect their own data despite routinely collected health data (RCHD) increasing in quality and breadth. Our aim is to quantify UK-based randomised controlled trials (RCTs) accessing RCHD for participant data, characterise how these data are used and thereby recommend how more trials could use RCHD.

METHODS

We conducted a systematic review of RCTs accessing RCHD from at least one registry in the UK between 2013 and 2018 for the purposes of informing or supplementing participant data. A list of all registries holding RCHD in the UK was compiled. In cases where registries published release registers, these were searched for RCTs accessing RCHD. Where no release register was available, registries were contacted to request a list of RCTs. For each identified RCT, information was collected from all publicly available sources (release registers, websites, protocol etc.). The search and data extraction were undertaken between January and May 2019.

RESULTS

We identified 160 RCTs accessing RCHD between 2013 and 2018 from a total of 22 registries; this corresponds to only a very small proportion of all UK RCTs (about 3%). RCTs accessing RCHD were generally large (median sample size 1590), commonly evaluating treatments for cancer or cardiovascular disease. Most of the included RCTs accessed RCHD from NHS Digital (68%), and the most frequently accessed datasets were mortality (76%) and hospital visits (55%). RCHD was used to inform the primary trial (82%) and long-term follow-up (57%). There was substantial variation in how RCTs used RCHD to inform participant outcome measures. A limitation was the lack of information and transparency from registries and RCTs with respect to which datasets have been accessed and for what purposes.

CONCLUSIONS

In the last five years, only a small minority of UK-based RCTs have accessed RCHD to inform participant data. We ask for improved accessibility, confirmed data quality and joined-up thinking between the registries and the regulatory authorities.

TRIAL REGISTRATION

PROSPERO CRD42019123088.

What are the implications of using individual and combined sources of routinely collected data to identify and characterise incident site-specific cancers? a concordance and validation study using linked English electronic health records data

OBJECTIVES

To describe the benefits and limitations of using individual and combinations of linked English electronic health data to identify incident cancers.

DESIGN AND SETTING

Our descriptive study uses linked English Clinical Practice Research Datalink primary care; cancer registration; hospitalisation and death registration data.

PARTICIPANTS AND MEASURES

We implemented case definitions to identify first site-specific cancers at the 20 most common sites, based on the first ever cancer diagnosis recorded in each individual or commonly used combination of data sources between 2000 and 2014. We calculated positive predictive values and sensitivities of each definition, compared with a gold standard algorithm that used information from all linked data sets to identify first cancers. We described completeness of grade and stage information in the cancer registration data set.

RESULTS

165 953 gold standard cancers were identified. Positive predictive values of all case definitions were ≥80% and ≥94% for the four most common cancers (breast, lung, colorectal and prostate). Sensitivity for case definitions that used cancer registration alone or in combination was ≥92% for the four most common cancers and ≥80% across all cancer sites except bladder cancer (65% using cancer registration alone). For case definitions using linked primary care, hospitalisation and death registration data, sensitivity was ≥89% for the four most common cancers, and ≥80% for all cancer sites except kidney (69%), oral cavity (76%) and ovarian cancer (78%). When primary care or hospitalisation data were used alone, sensitivities were generally lower and diagnosis dates were delayed. Completeness of staging data in cancer registration data was high from 2012 (minimum 76.0% in 2012 and 86.4% in 2014 for the four most common cancers).

CONCLUSIONS

Ascertainment of incident cancers was good when using cancer registration data alone or in combination with other data sets, and for the majority of cancers when using a combination of primary care, hospitalisation and death registration data.

Comparison of the Accuracy and Completeness of Records of Serious Vascular Events in Routinely Collected Data vs Clinical Trial-Adjudicated Direct Follow-up Data in the UK: Secondary Analysis of the ASCEND Randomized Clinical Trial

Importance

Routinely collected data could substantially decrease the cost of conducting trials.

Objective

To assess the accuracy and completeness of UK routine data for ascertaining serious vascular events (SVEs) compared with adjudicated follow-up data.

Design, Setting, and Participants

This was a secondary analysis of a randomized clinical trial. From June 24, 2005, to July 28, 2011, the ASCEND (A Study of Cardiovascular Events in Diabetes) primary prevention trial used mail-based methods to randomize people with diabetes without evidence of atherosclerotic vascular disease using a 2 × 2 factorial design to aspirin and/or ω-fatty acids vs matching placebo in the UK. Direct participant mail-based follow-up was the main source of outcome data, with more than 90% of the primary outcome events undergoing adjudication. Follow-up was completed on July 31, 2017. In parallel, more than 99% of participants were linked to routinely collected hospital admission and death registry data (ie, routine data), enabling post hoc randomized comparisons of different sources of outcome data (conducted from September 1, 2018, to October 1, 2021).

Interventions

Random allocation to 100 mg of aspirin once daily vs matching placebo and separately to 1 g of ω-3 fatty acids once daily vs placebo.

Main Outcomes and Measures

The primary outcome consisted of SVEs (a composite of nonfatal myocardial infarction, ischemic stroke, transient ischemic attack [TIA], or vascular death, excluding hemorrhagic stroke).

Results

A total of 15 480 participants were randomized (mean [SD] age, 63 [9] years; 9684 [62.6%] men) and followed up for a mean (SD) of 7.4 (1.8) years. For SVEs, agreement between adjudicated direct follow-up and routine data sources was strong (1401 vs 1127 events; κ = 0.78 [95% CI, 0.76-0.80]; sensitivity, 72.0% [95% CI, 69.7%-74.4%]; specificity, 99.2% [95% CI, 99.0%-99.3%]), and sensitivity improved for SVEs excluding transient ischemic attack (1129 vs 1026 events; sensitivity, 80.6% [95% CI, 78.3%-82.9%]). Rate ratios for the aspirin-randomized comparison for adjudicated direct follow-up vs follow-up solely through routine data alone were 0.88 (95% CI, 0.79-0.97) vs 0.91 (95% CI, 0.81-1.02) for the primary outcome and 0.92 (95% CI, 0.82-1.03) vs 0.91 (95% CI, 0.80-1.02) for SVEs excluding TIA. Results were similar for the ω-3 fatty acid comparison, and adjudication did not seem to markedly change rate ratios.

Conclusions and Relevance

Post hoc analyses of the ASCEND trial suggest that routinely collected hospital admission and death registry data in the UK could be used as the sole method of follow-up for myocardial infarction, ischemic stroke resulting in hospitalization, vascular death, and arterial revascularization in primary prevention cardiovascular trials, without the need for verification by clinical adjudication.

Prediction and visualization of acute kidney injury in intensive care unit using one-dimensional convolutional neural networks based on routinely collected data

BACKGROUND

Acute kidney injury (AKI) occurs frequently in in-hospital patients, especially in the intensive care unit (ICU), due to various etiologies including septic shock. It is clinically important to identify high-risk patients at an early stage and perform the appropriate intervention.

METHODS

We proposed a system to predict AKI using one-dimensional convolutional neural networks (1D-CNN) with the real-time calculation of the probability of developing AKI, along with the visualization of the rationale behind prediction using score-weighted class activation mapping and guided backpropagation. The system was applied to predicting developing AKI based on the KDIGO guideline in time windows of 24 to 48 h using data of 0 to 24 h after admission to ICU.

RESULTS

The comparison result of multiple algorithms modeling time series data indicated that the proposed 1D-CNN model achieved higher performance compared to the other models, with the mean area under the receiver operating characteristic curve of 0.742 ± 0.010 for predicting stage 1, and 0.844 ± 0.029 for stage 2 AKI using the input of the vital signs, the demographic information, and serum creatinine values. The visualization results suggested the reasonable interpretation that time points with higher respiratory rate, lower blood pressure, as well as lower SpO2, had higher attention in terms of predicting AKI, and thus important for prediction.

CONCLUSIONS

We presumed the proposed system's potential usefulness as it could be applied and transferred to almost any ICU setting that stored the time series data corresponding to vital signs.

Trends in percentages of gestational diabetes mellitus attributable to overweight, obesity, and morbid obesity in regional Victoria: an eight-year population-based panel study

BACKGROUND

Gestational diabetes mellitus (GDM) is the fastest growing type of diabetes in Australia with rates trebling over the past decades partially explained by rising obesity rates and maternal age among childbearing women. Percentage of GDM attributable to obesity has been documented, mostly focusing on metropolitan populations. In parts of regional (areas outside capital cities) and rural Australia where overweight, obesity and morbid obesity are more prevalent, intertwined with socioeconomic disadvantage and higher migrant communities, trends over time in adjusted percentages of GDM attributed to obesity are unknown.

METHODS

In this population-based retrospective panel study, women, without pre-existing diabetes, delivering singletons between 2010 and 2017 in a tertiary regional hospital that serves 26% of Victoria's 6.5 million Australian population were eligible for inclusion. Secular trends in GDM by body mass index (BMI) and age were evaluated. The percentage of GDM that would have been prevented each year with the elimination of overweight or obesity was estimated using risk-adjusted regression-based population attributable fractions (AFp). Trends in the AFp over time were tested using the augmented Dickey-Fuller test.

RESULTS

Overall 7348 women, contributing to 10,028 births were included. The age of expecting mothers, their BMI, proportion of women born overseas, and GDM incidence significantly rose over time with GDM rising from 3.5% in 2010 to 13.7% in 2017, p <  0.001, increasing in all BMI categories. The incidence was consistently highest among women with obesity (13.8%) and morbid obesity (21.6%). However, the highest relative increase was among women with BMI < 25 kg/m2, rising from 1.4% in 2010 to 7.0% in 2017. Adjusting for age, country of birth, socioeconomic status, comorbidities, antenatal and intrapartum factors, an estimated 8.6% (confidence interval (CI) 6.1-11.0%), 15.6% (95% CI 12.2-19.0%), and 19.5% (95% CI 15.3-23.6%) of GDM would have been prevented by eliminating maternal overweight, obesity, and morbid obesity, respectively. However, despite the rise in obesity over time, percentages of GDM attributable to overweight, obesity, and morbid obesity significantly dropped over time. Scenario analyses supported these findings.

CONCLUSIONS

Besides increasing prevalence of obesity over time, this study suggests that GDM risk factors, other than obesity, are also increasing over time.

Big Data and diabetes: the applications of Big Data for diabetes care now and in the future

We review current applications of Big Data in diabetes care and consider the future potential by carrying out a scoping study of the academic literature on Big Data and diabetes care. Healthcare data are being produced at ever-increasing rates, and this information has the potential to transform the provision of diabetes care. Big Data is beginning to have an impact on diabetes care through data research. The use of Big Data for routine clinical care is still a future application. Vast amounts of healthcare data are already being produced, and the key is harnessing these to produce actionable insights. Considerable development work is required to achieve these goals.

High-dimensional propensity scores for empirical covariate selection in secondary database studies: Planning, implementation, and reporting

Real-world evidence used for regulatory, payer, and clinical decision-making requires principled epidemiology in design and analysis, applying methods to minimize confounding given the lack of randomization. One technique to deal with potential confounding is propensity score (PS) analysis, which allows for the adjustment for measured preexposure covariates. Since its first publication in 2009, the high-dimensional propensity score (hdPS) method has emerged as an approach that extends traditional PS covariate selection to include large numbers of covariates that may reduce confounding bias in the analysis of healthcare databases. hdPS is an automated, data-driven analytic approach for covariate selection that empirically identifies preexposure variables and proxies to include in the PS model. This article provides an overview of the hdPS approach and recommendations on the planning, implementation, and reporting of hdPS used for causal treatment-effect estimations in longitudinal healthcare databases. We supply a checklist with key considerations as a supportive decision tool to aid investigators in the implementation and transparent reporting of hdPS techniques, and to aid decision-makers unfamiliar with hdPS in the understanding and interpretation of studies employing this approach. This article is endorsed by the International Society for Pharmacoepidemiology.

Understanding challenges of using routinely collected health data to address clinical care gaps: a case study in Alberta, Canada

High-quality data are fundamental to healthcare research, future applications of artificial intelligence and advancing healthcare delivery and outcomes through a learning health system. Although routinely collected administrative health and electronic medical record data are rich sources of information, they have significant limitations. Through four example projects from the Physician Learning Program in Edmonton, Alberta, Canada, we illustrate barriers to using routinely collected health data to conduct research and engage in clinical quality improvement. These include challenges with data availability for variables of clinical interest, data completeness within a clinical visit, missing and duplicate visits, and variability of data capture systems. We make four recommendations that highlight the need for increased clinical engagement to improve the collection and coding of routinely collected data. Advancing the quality and usability of health systems data will support the continuous quality improvement needed to achieve the quintuple aim.

Uptake of infant and preschool immunisations in Scotland and England during the COVID-19 pandemic: An observational study of routinely collected data

BACKGROUND

In 2020, the SARS-CoV-2 (COVID-19) pandemic and lockdown control measures threatened to disrupt routine childhood immunisation programmes with early reports suggesting uptake would fall. In response, public health bodies in Scotland and England collected national data for childhood immunisations on a weekly or monthly basis to allow for rapid analysis of trends. The aim of this study was to use these data to assess the impact of different phases of the pandemic on infant and preschool immunisation uptake rates.

METHODS AND FINDINGS

We conducted an observational study using routinely collected data for the year prior to the pandemic (2019) and immediately before (22 January to March 2020), during (23 March to 26 July), and after (27 July to 4 October) the first UK "lockdown". Data were obtained for Scotland from the Public Health Scotland "COVID19 wider impacts on the health care system" dashboard and for England from ImmForm. Five vaccinations delivered at different ages were evaluated; 3 doses of "6-in-1" diphtheria, tetanus, pertussis, polio, Haemophilus influenzae type b, and hepatitis B vaccine (DTaP/IPV/Hib/HepB) and 2 doses of measles, mumps, and rubella (MMR) vaccine. This represented 439,754 invitations to be vaccinated in Scotland and 4.1 million for England. Uptake during the 2020 periods was compared to the previous year (2019) using binary logistic regression analysis. For Scotland, uptake within 4 weeks of a child becoming eligible by age was analysed along with geographical region and indices of deprivation. For Scotland and England, we assessed whether immunisations were up-to-date at approximately 6 months (all doses 6-in-1) and 16 to 18 months (first MMR) of age. We found that uptake within 4 weeks of eligibility in Scotland for all the 5 vaccines was higher during lockdown than in 2019. Differences ranged from 1.3% for first dose 6-in-1 vaccine (95.3 versus 94%, odds ratio [OR] compared to 2019 1.28, 95% confidence intervals [CIs] 1.18 to 1.39) to 14.3% for second MMR dose (66.1 versus 51.8%, OR compared to 2019 1.8, 95% CI 1.74 to 1.87). Significant increases in uptake were seen across all deprivation levels. In England, fewer children due to receive their immunisations during the lockdown period were up to date at 6 months (6-in-1) or 18 months (first dose MMR). The fall in percentage uptake ranged from 0.5% for first 6-in-1 (95.8 versus 96.3%, OR compared to 2019 0.89, 95% CI 0.86- to 0.91) to 2.1% for third 6-in-1 (86.6 versus 88.7%, OR compared to 2019 0.82, 95% CI 0.81 to 0.83). The use of routinely collected data used in this study was a limiting factor as detailed information on potential confounding factors were not available and we were unable to eliminate the possibility of seasonal trends in immunisation uptake.

CONCLUSIONS

In this study, we observed that the national lockdown in Scotland was associated with an increase in timely childhood immunisation uptake; however, in England, uptake fell slightly. Reasons for the improved uptake in Scotland may include active measures taken to promote immunisation at local and national levels during this period and should be explored further. Promoting immunisation uptake and addressing potential vaccine hesitancy is particularly important given the ongoing pandemic and COVID-19 vaccination campaigns.

Dementia prevalence estimation among the main ethnic groups in New Zealand: a population-based descriptive study of routinely collected health data

OBJECTIVE

Estimates of dementia prevalence in New Zealand (NZ) have previously been extrapolated from limited Australasian studies, which may be neither accurate nor reflect NZ's unique population and diverse ethnic groups. This study used routinely collected health data to estimate the 1-year period prevalence for diagnosed dementia for each of the 4 years between July 2016 and June 2020 in the age 60+ and age 80+ populations and for the four main ethnic groups.

DESIGN

A population-based descriptive study.

SETTING

Seven national health data sets within the NZ Integrated Data Infrastructure (IDI) were linked. Diagnosed dementia prevalence for each year was calculated using the IDI age 60+ and age 80+ populations as the denominator and also age-sex standardised to allow comparison across ethnic groups.

PARTICIPANTS

Diagnosed dementia individuals in the health datasets were identified by diagnostic or medication codes used in each of the data sets with deduplication of those who appeared in more than one data set.

RESULTS

The crude diagnosed dementia prevalence was 3.8%-4.0% in the age 60+ population and 13.7%-14.4% in the age 80+ population across the four study years. Dementia prevalence age-sex standardised to the IDI population in the last study period of 2019-2020 was 5.4% for Māori, 6.3% for Pacific Islander, 3.7% for European and 3.4% for Asian in the age 60+ population, and 17.5% for Māori, 22.2% for Pacific Islander, 13.6% for European and 13.5% for Asian in the age 80+ population.

CONCLUSIONS

This study provides the best estimate to date for dementia prevalence in NZ but is limited to those people who were identified as having dementia based on data from the seven included data sets. The findings suggest that diagnosed dementia prevalence is higher in Māori and Pacific Islanders. A nationwide NZ community-based dementia prevalence study is much needed to confirm the findings of this study.

Days alive and at home after hip fracture: a cross-sectional validation of a patient-centred outcome measure using routinely collected data

BACKGROUND

Days alive and at home (DAH) is a patient centered outcome measureable in routinely collected health data. The validity and minimally important difference (MID) in hip fracture have not been evaluated.

OBJECTIVE

We assessed construct and predictive validity and estimated a MID for the patient-centred outcome of DAH after hip fracture admission.

METHODS

This is a cross-sectional observational study using linked health administrative data in Ontario, Canada. DAH was calculated as the number of days alive within 90 days of admission minus the number of days hospitalised or institutionalised. All hospital admissions (2012-2018) for hip fracture in adults aged >50 years were included. Construct validity analyses used Bayesian quantile regression to estimate the associations of postulated patient, admission and process-related variables with DAH. The predictive validity assessed was the correlation of DAH in 90 days with the value from 91 to 365 days; and the association and discrimination of DAH in 90 days predicting subsequent mortality. MID was estimated by averaging distribution-based and clinical anchor-based estimates.

RESULTS

We identified 63 778 patients with hip fracture. The median number of DAH was 43 (range 0-87). In the 90 days after admission, 8050 (12.6%) people died; a further 6366 (10.0%) died from days 91 to 365. Associations between patient-level and admission-level factors with the median DAH (lower with greater age, frailty and comorbidity, lower if admitted to intensive care or having had a complication) supported construct validity. DAH in 90 days after admission was strongly correlated with DAH in 365 days after admission (r=0.922). An 11-day MID was estimated.

CONCLUSION

DAH has face, construct and predictive validity as a patient-centred outcome in patients with hip fracture, with an estimated MID of 11 days. Future research is required to include direct patient perspectives in confirming MID.

Surgical or percutaneous coronary revascularization for heart failure: an in silico model using routinely collected health data to emulate a clinical trial

AIMS

The choice of revascularization with coronary artery bypass grafting (CABG) vs. percutaneous coronary intervention (PCI) in people with ischaemic left ventricular dysfunction is not guided by high-quality evidence.

METHODS AND RESULTS

A trial of CABG vs. PCI in people with heart failure (HF) was modelled in silico using routinely collected healthcare data. The in silico trial cohort was selected by matching the target trial cohort, identified from Hospital Episode Statistics in England, with individual patient data from the Surgical Treatment for Ischemic Heart Failure (STICH) trial. Allocation to CABG vs. complex PCI demonstrated random variation across administrative regions in England and was a valid statistical instrument. The primary outcome was 5-year all-cause mortality or cardiovascular hospitalization. Instrumental variable analysis (IVA) was used for the primary analysis. Results were expressed as average treatment effects (ATEs) with 95 confidence intervals (CIs). The target population included 13 519 HF patients undergoing CABG or complex PCI between April 2009 and March 2015. After matching, the emulated trial cohort included 2046 patients. The unadjusted primary outcome rate was 51.1 in the CABG group and 70.0 in the PCI group. IVA of the emulated cohort showed that CABG was associated with a lower risk of the primary outcome (ATE 16.2, 95 CI 20.6 to 11.8), with comparable estimates in the unmatched target population (ATE 15.5, 95 CI 17.5 to 13.5).

CONCLUSION

In people with HF, in silico modelling suggests that CABG is associated with fewer deaths or cardiovascular hospitalizations at 5 years vs. complex PCI. A pragmatic clinical trial is needed to test this hypothesis and this trial would be feasible.

Using the Robson classification to assess caesarean section rates in Brazil: an observational study of more than 24 million births from 2011 to 2017

BACKGROUND

Applying the Robson classification to all births in Brazil, the objectives of our study were to estimate the rates of caesarean section delivery, assess the extent to which caesarean sections were clinically indicated, and identify variation across socioeconomic groups.

METHODS

We conducted a population-based study using routine records of the Live Births Information System in Brazil from January 1, 2011, to December 31, 2017. We calculated the relative size of each Robson group; the caesarean section rate; and the contribution to the overall caesarean section rate. We categorised Brazilian municipalities using the Human Development Index to explore caesarean section rates further. We estimated the time trend in caesarean section rates.

RESULTS

The rate of caesarean sections was higher in older and more educated women. Prelabour caesarean sections accounted for more than 54 % of all caesarean deliveries. Women with a previous caesarean section (Group 5) made up the largest group (21.7 %). Groups 6-9, for whom caesarean sections would be indicated in most cases, all had caesarean section rates above 82 %, as did Group 5. The caesarean section rates were higher in municipalities with a higher HDI. The general Brazilian caesarean section rate remained stable during the study period.

CONCLUSIONS

Brazil is a country with one of the world's highest caesarean section rates. This nationwide population-based study provides the evidence needed to inform efforts to improve the provision of clinically indicated caesarean sections. Our results showed that caesarean section rates were lower among lower socioeconomic groups even when clinically indicated, suggesting sub-optimal access to surgical care.

Patient and Public Willingness to Share Personal Health Data for Third-Party or Secondary Uses: Systematic Review

BACKGROUND

International advances in information communication, eHealth, and other digital health technologies have led to significant expansions in the collection and analysis of personal health data. However, following a series of high-profile data sharing scandals and the emergence of COVID-19, critical exploration of public willingness to share personal health data remains limited, particularly for third-party or secondary uses.

OBJECTIVE

This systematic review aims to explore factors that affect public willingness to share personal health data for third-party or secondary uses.

METHODS

A systematic search of 6 databases (MEDLINE, Embase, PsycINFO, CINAHL, Scopus, and SocINDEX) was conducted with review findings analyzed using inductive-thematic analysis and synthesized using a narrative approach.

RESULTS

Of the 13,949 papers identified, 135 were included. Factors most commonly identified as a barrier to data sharing from a public perspective included data privacy, security, and management concerns. Other factors found to influence willingness to share personal health data included the type of data being collected (ie, perceived sensitivity); the type of user requesting their data to be shared, including their perceived motivation, profit prioritization, and ability to directly impact patient care; trust in the data user, as well as in associated processes, often established through individual choice and control over what data are shared with whom, when, and for how long, supported by appropriate models of dynamic consent; the presence of a feedback loop; and clearly articulated benefits or issue relevance including valued incentivization and compensation at both an individual and collective or societal level.

CONCLUSIONS

There is general, yet conditional public support for sharing personal health data for third-party or secondary use. Clarity, transparency, and individual control over who has access to what data, when, and for how long are widely regarded as essential prerequisites for public data sharing support. Individual levels of control and choice need to operate within the auspices of assured data privacy and security processes, underpinned by dynamic and responsive models of consent that prioritize individual or collective benefits over and above commercial gain. Failure to understand, design, and refine data sharing approaches in response to changeable patient preferences will only jeopardize the tangible benefits of data sharing practices being fully realized.

Survivorship of the dual-mobility construct in elective primary total hip replacement: a systematic review and meta-analysis including registry data

INTRODUCTION

Dislocation is a common complication associated with total hip replacement (THR). Dual-mobility constructs (DMC-THR) may be used in high-risk patients and have design features that may reduce the risk of dislocation. We aimed to report overall pooled estimates of all-cause construct survival for elective primary DMC-THR. Secondary outcomes included unadjusted dislocation rate, revision for instability, infection and fracture.

METHODS

MEDLINE, EMBASE, Web of Science, Cochrane Library and National Joint Registry reports were systematically searched (CRD42020189664). Studies reporting revision (all-cause) survival estimates and confidence intervals by brand and construct including DMC bearings were included. A meta-analysis was performed weighting series by the standard error.

RESULTS

Thirty-seven studies reporting 39 case series were identified; nine (10,494 DMC-THR) were included. Fourteen series (23,020 DMC-THR) from five national registries were included. Pooled case series data for all-cause construct survival was 99.7% (95% CI 99.5-100) at 5 years, 95.7% (95% CI 94.9-96.5) at 10 years, 96.1% (95% CI 91.8-100) at 15 years and 77% (95% CI 74.4-82.0) at 20 years. Pooled joint registry data showed an all-cause construct survivorship of 97.8% (95% CI 97.3-98.4) at 5 years and 96.3% (95% CI 95.6-96.9) at 10 years.

CONCLUSIONS

Survivorship of DMC-THR in primary THR is acceptable according to the national revision benchmark published by National Institute for Clinical Excellence (NICE).

Early influenza vaccine effectiveness estimates using routinely collected data, Alberta, Canada, 2023/24 season

Timely and precise influenza vaccine effectiveness (VE) estimates are needed to guide public health messaging and impact vaccine uptake immediately. Using routinely collected laboratory, vaccination and health administrative data from Alberta, Canada, we estimated influenza VE against infection for the 2023/24 season on a near real-time basis, to late December, at 61% (95% CI: 58-64) against influenza A(H1N1), 49% (95% CI: 28-63) against influenza A(H3N2) and 75% (95% CI: 58-85) against influenza B.

How many days are needed? Measurement reliability of wearable device data to assess physical activity

INTRODUCTION/PURPOSE

Physical activity studies often utilize wearable devices to measure participants' habitual activity levels by averaging values across several valid observation days. These studies face competing demands-available resources and the burden to study participants must be balanced with the goal to obtain reliable measurements of a person's longer-term average. Information about the number of valid observation days required to reliably measure targeted metrics of habitual activity is required to inform study design.

METHODS

To date, the number of days required to achieve a desired level of aggregate long-term reliability (typically 0.80) has often been estimated by applying the Spearman-Brown Prophecy formula to short-term test-retest reliability data from studies with single, relatively brief observation windows. Our work, in contrast, utilizes a resampling-based approach to quantify the long-term test-retest reliability of aggregate measures of activity in a cohort of 79 participants who were asked to wear a FitBit Flex every day for approximately one year.

RESULTS

The conventional approach can produce reliability estimates that substantially overestimate the actual test-retest reliability. Six or more valid days of observation for each participant appear necessary to obtain 0.80 reliability for the average amount of time spent in light physical activity; 8 and 10 valid days are needed for sedentary time and moderate/vigorous activity respectively.

CONCLUSION

Protocols that result in 7-10 valid observation days for each participant may be needed to obtain reliable measurements of key physical activity metrics.