Ontology to identify pregnant women in electronic health records: primary care sentinel network database study

Phenotype execution and modeling architecture to support disease surveillance and real-world evidence studies: English sentinel network evaluation

Objective

To evaluate Phenotype Execution and Modelling Architecture (PhEMA), to express sharable phenotypes using Clinical Quality Language (CQL) and intensional Systematised Nomenclature of Medicine (SNOMED) Clinical Terms (CT) Fast Healthcare Interoperability Resources (FHIR) valuesets, for exemplar chronic disease, sociodemographic risk factor, and surveillance phenotypes.

Method

We curated 3 phenotypes: Type 2 diabetes mellitus (T2DM), excessive alcohol use, and incident influenza-like illness (ILI) using CQL to define clinical and administrative logic. We defined our phenotypes with valuesets, using SNOMED's hierarchy and expression constraint language, and CQL, combining valuesets and adding temporal elements where needed. We compared the count of cases found using PhEMA with our existing approach using convenience datasets. We assessed our new approach against published desiderata for phenotypes.

Results

The T2DM phenotype could be defined as 2 intensionally defined SNOMED valuesets and a CQL script. It increased the prevalence from 7.2% to 7.3%. Excess alcohol phenotype was defined by valuesets that added qualitative clinical terms to the quantitative conceptual definitions we currently use; this change increased prevalence by 58%, from 1.2% to 1.9%. We created an ILI valueset with SNOMED concepts, adding a temporal element using CQL to differentiate new episodes. This increased the weekly incidence in our convenience sample (weeks 26-38) from 0.95 cases to 1.11 cases per 100 000 people.

Conclusions

Phenotypes for surveillance and research can be described fully and comprehensibly using CQL and intensional FHIR valuesets. Our use case phenotypes identified a greater number of cases, whilst anticipated from excessive alcohol this was not for our other variable. This may have been due to our use of SNOMED CT hierarchy. Our new process fulfilled a greater number of phenotype desiderata than the one that we had used previously, mostly in the modeling domain. More work is needed to implement that sharing and warehousing domains.

Hepatitis B virus infection in general practice across England: An analysis of the Royal College of General Practitioners Research and Surveillance Centre real-world database

OBJECTIVES

We analysed hepatitis B surface antigen (HBsAg) screening and seropositivity within a network of 419 general practices representative of all regions of England.

METHODS

Information was extracted using pseudonymised registration data. Predictors of HBsAg seropositivity were explored in models that considered age, gender, ethnicity, time at the current practice, practice location and associated deprivation index, and presence of nationally endorsed screen indicators including pregnancy, men who have sex with men (MSM), history of injecting drug use (IDU), close HBV contact or imprisonment, and diagnosis of blood-borne or sexually transmitted infections.

RESULTS

Among 6,975,119 individuals, 192,639 (2.8%) had a screening record, including 3.6- 38.6% of those with a screen indicator, and 8065 (0.12%) had a seropositive record. The odds of seropositivity were highest in London, in the most deprived neighbourhoods, among minority ethnic groups, and in people with screen indicators. Seroprevalence exceeded 1% in people from high prevalence countries, MSM, close HBV contacts, and people with a history of IDU or HIV, HCV, or syphilis diagnosis. Overall, 1989/8065 (24.7%) had a recorded referral to specialist hepatitis care.

CONCLUSIONS

In England, HBV infection is associated with poverty. There are unrealised opportunities to promote access to diagnosis and care for those affected.

Temporal Events Detector for Pregnancy Care (TED-PC): A rule-based algorithm to infer gestational age and delivery date from electronic health records of pregnant women with and without COVID-19

OBJECTIVE

Identifying the time of SARS-CoV-2 viral infection relative to specific gestational weeks is critical for delineating the role of viral infection timing in adverse pregnancy outcomes. However, this task is difficult when it comes to Electronic Health Records (EHR). In combating the COVID-19 pandemic for maternal health, we sought to develop and validate a clinical information extraction algorithm to detect the time of clinical events relative to gestational weeks.

MATERIALS AND METHODS

We used EHR from the National COVID Cohort Collaborative (N3C), in which the EHR are normalized by the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). We performed EHR phenotyping, resulting in 270,897 pregnant women (June 1st, 2018 to May 31st, 2021). We developed a rule-based algorithm and performed a multi-level evaluation to test content validity and clinical validity, and extreme length of gestation (<150 or >300).

RESULTS

The algorithm identified 296,194 pregnancies (16,659 COVID-19, 174,744 without COVID-19) in 270,897 pregnant women. For inferring gestational age, 95% cases (n = 40) have moderate-high accuracy (Cohen's Kappa = 0.62); 100% cases (n = 40) have moderate-high granularity of temporal information (Cohen's Kappa = 1). For inferring delivery dates, the accuracy is 100% (Cohen's Kappa = 1). The accuracy of gestational age detection for the extreme length of gestation is 93.3% (Cohen's Kappa = 1). Mothers with COVID-19 showed higher prevalence in obesity or overweight (35.1% vs. 29.5%), diabetes (17.8% vs. 17.0%), chronic obstructive pulmonary disease (0.2% vs. 0.1%), respiratory distress syndrome or acute respiratory failure (1.8% vs. 0.2%).

DISCUSSION

We explored the characteristics of pregnant women by different gestational weeks of SARS-CoV-2 infection with our algorithm. TED-PC is the first to infer the exact gestational week linked with every clinical event from EHR and detect the timing of SARS-CoV-2 infection in pregnant women.

CONCLUSION

The algorithm shows excellent clinical validity in inferring gestational age and delivery dates, which supports multiple EHR cohorts on N3C studying the impact of COVID-19 on pregnancy.

Medication history-wide association studies for pharmacovigilance of pregnant patients

Background

Systematic exclusion of pregnant people from interventional clinical trials has created a public health emergency for millions of patients through a dearth of robust safety data for common drugs.

Methods

We harnessed an enterprise collection of 2.8 M electronic health records (EHRs) from routine care, leveraging data linkages between mothers and their babies to detect drug safety signals in this population at full scale. Our mixed-methods signal detection approach stimulates new hypotheses for post-marketing surveillance agnostically of both drugs and diseases—by identifying 1,054 drugs historically prescribed to pregnant patients; developing a quantitative, medication history-wide association study; and integrating a qualitative evidence synthesis platform using expert clinician review for integration of biomedical specificity—to test the effects of maternal exposure to diverse drugs on the incidence of neurodevelopmental defects in their children.

Results

We replicated known teratogenic risks and existing knowledge on drug structure-related teratogenicity; we also highlight 5 common drug classes for which we believe this work warrants updated assessment of their safety.

Conclusion

Here, we present roots of an agile framework to guide enhanced medication regulations, as well as the ontological and analytical limitations that currently restrict the integration of real-world data into drug safety management during pregnancy. This research is not a replacement for inclusion of pregnant people in prospective clinical studies, but it presents a tractable team science approach to evaluating the utility of EHRs for new regulatory review programs—towards improving the delicate equipoise of accuracy and ethics in assessing drug safety in pregnancy.

Investigating the uptake, effectiveness and safety of COVID-19 vaccines: protocol for an observational study using linked UK national data

INTRODUCTION

The novel coronavirus SARS-CoV-2, which emerged in December 2019, has caused millions of deaths and severe illness worldwide. Numerous vaccines are currently under development of which a few have now been authorised for population-level administration by several countries. As of 20 September 2021, over 48 million people have received their first vaccine dose and over 44 million people have received their second vaccine dose across the UK. We aim to assess the uptake rates, effectiveness, and safety of all currently approved COVID-19 vaccines in the UK.

METHODS AND ANALYSIS

We will use prospective cohort study designs to assess vaccine uptake, effectiveness and safety against clinical outcomes and deaths. Test-negative case-control study design will be used to assess vaccine effectiveness (VE) against laboratory confirmed SARS-CoV-2 infection. Self-controlled case series and retrospective cohort study designs will be carried out to assess vaccine safety against mild-to-moderate and severe adverse events, respectively. Individual-level pseudonymised data from primary care, secondary care, laboratory test and death records will be linked and analysed in secure research environments in each UK nation. Univariate and multivariate logistic regression models will be carried out to estimate vaccine uptake levels in relation to various population characteristics. VE estimates against laboratory confirmed SARS-CoV-2 infection will be generated using a generalised additive logistic model. Time-dependent Cox models will be used to estimate the VE against clinical outcomes and deaths. The safety of the vaccines will be assessed using logistic regression models with an offset for the length of the risk period. Where possible, data will be meta-analysed across the UK nations.

ETHICS AND DISSEMINATION

We obtained approvals from the National Research Ethics Service Committee, Southeast Scotland 02 (12/SS/0201), the Secure Anonymised Information Linkage independent Information Governance Review Panel project number 0911. Concerning English data, University of Oxford is compliant with the General Data Protection Regulation and the National Health Service (NHS) Digital Data Security and Protection Policy. This is an approved study (Integrated Research Application ID 301740, Health Research Authority (HRA) Research Ethics Committee 21/HRA/2786). The Oxford-Royal College of General Practitioners Clinical Informatics Digital Hub meets NHS Digital's Data Security and Protection Toolkit requirements. In Northern Ireland, the project was approved by the Honest Broker Governance Board, project number 0064. Findings will be made available to national policy-makers, presented at conferences and published in peer-reviewed journals.

Valproate prescription to women of childbearing age in English primary care: repeated cross-sectional analyses and retrospective cohort study

Background

Valproate is a teratogenic drug that should be avoided during the preconception period and pregnancy. The aim was to explore general practitioners’ (GPs) prescription patterns over time, describe trends, and explore inter-practice variation within primary care.

Methods

We identified women of childbearing age (12–46 years old) in the Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) sentinel network. We performed repeated cross-sectional analyses from 2004 to 2018 to determine rates of prescription and a retrospective cohort estimated the prevalence of use of valproate during pregnancy.

Results

In 2004, 0.31% (95% Confidence Interval (95%CI):0.18 to 0.44%) women were prescribed valproate, decreasing to 0.16% (95%CI:0.07 to 0.24%) by 2018. Among women with epilepsy, the rate fell from 15.2% (95%CI:14.4 to 16.0%) to 8.8% (95% CI:8.2 to 9.7%) over the same period. In 2018, almost two thirds (62.2%) of women who were prescribed valproate had epilepsy only, whereas bipolar disorder and migraine accounted for 15.8% and 7.4% respectively. Contraceptive prescriptions did not increase over time, and only in 2018 was there greater odds of being prescribed contraception (OR 1.41, 95%CI:1.08 to 1.45). Just under a fifth (19.7%) of women were prescribed valproate during their pregnancy; two out of three of these pregnancies were preceded by folic acid prescription (5 mg). While some practices reduced their rate of valproate prescription, others did not.

Conclusions

Regulatory guidelines have changed GPs' prescription patterns in women of childbearing potential for valproate but not for contraception. Further research is needed to identify the barriers of GPs and women of childbearing potential to undertaking contraception.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-021-04351-x.

Household presentation of influenza and acute respiratory illnesses to a primary care sentinel network: retrospective database studies (2013-2018)

Background

Direct observation of the household spread of influenza and respiratory infections is limited; much of our understanding comes from mathematical models. The study aims to determine household incidence of influenza-like illness (ILI), lower (LRTI) and upper (URTI) respiratory infections within a primary care routine data and identify factors associated with the diseases’ incidence.

Methods

We conducted two five-year retrospective analyses of influenza-like illness (ILI), lower (LRTI) and upper (URTI) respiratory infections using the England Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) primary care sentinel network database; a cross-sectional study reporting incident rate ratio (IRR) from a negative binomial model and a retrospective cohort study, using a shared gamma frailty survival model, reporting hazard ratios (HR). We reported the following household characteristics: children < 5 years old, each extra household member, gender, ethnicity (reference white), chronic disease, pregnancy, and rurality.

Results

The IRR where there was a child < 5 years were 1·62 (1·38–1·89, p < 0·0001), 2·40 (2.04–2.83, p < 0·0001) and 4·46 (3.79–5.255, p < 0·0001) for ILI, LRTI and URTI respectively. IRR also increased with household size, rurality and presentations and by female gender, compared to male. Household incidence of URTI and LRTI changed little between years whereas influenza did and were greater in years with lower vaccine effectiveness.

The HR where there was a child < 5 years were 2·34 (95%CI 1·88–2·90, p < 0·0001), 2·97 (95%CI 2·76–3·2, p < 0·0001) and 10·32 (95%CI 10.04–10.62, p < 0·0001) for ILI, LRTI and URTI respectively. HR were increased with female gender, rurality, and increasing household size.

Conclusions

Patterns of household incidence can be measured from routine data and may provide insights for the modelling of disease transmission and public health policy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-020-09790-3.

COVID-19 Surveillance in a Primary Care Sentinel Network: In-Pandemic Development of an Application Ontology

Background

Creating an ontology for COVID-19 surveillance should help ensure transparency and consistency. Ontologies formalize conceptualizations at either the domain or application level. Application ontologies cross domains and are specified through testable use cases. Our use case was an extension of the role of the Oxford Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) to monitor the current pandemic and become an in-pandemic research platform.

Objective

This study aimed to develop an application ontology for COVID-19 that can be deployed across the various use-case domains of the RCGP RSC research and surveillance activities.

Methods

We described our domain-specific use case. The actor was the RCGP RSC sentinel network, the system was the course of the COVID-19 pandemic, and the outcomes were the spread and effect of mitigation measures. We used our established 3-step method to develop the ontology, separating ontological concept development from code mapping and data extract validation. We developed a coding system–independent COVID-19 case identification algorithm. As there were no gold-standard pandemic surveillance ontologies, we conducted a rapid Delphi consensus exercise through the International Medical Informatics Association Primary Health Care Informatics working group and extended networks.

Results

Our use-case domains included primary care, public health, virology, clinical research, and clinical informatics. Our ontology supported (1) case identification, microbiological sampling, and health outcomes at an individual practice and at the national level; (2) feedback through a dashboard; (3) a national observatory; (4) regular updates for Public Health England; and (5) transformation of a sentinel network into a trial platform. We have identified a total of 19,115 people with a definite COVID-19 status, 5226 probable cases, and 74,293 people with possible COVID-19, within the RCGP RSC network (N=5,370,225).

Conclusions

The underpinning structure of our ontological approach has coped with multiple clinical coding challenges. At a time when there is uncertainty about international comparisons, clarity about the basis on which case definitions and outcomes are made from routine data is essential.

Emergence of a Novel Coronavirus (COVID-19): Protocol for Extending Surveillance Used by the Royal College of General Practitioners Research and Surveillance Centre and Public Health England

Background

The Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) and Public Health England (PHE) have successfully worked together on the surveillance of influenza and other infectious diseases for over 50 years, including three previous pandemics. With the emergence of the international outbreak of the coronavirus infection (COVID-19), a UK national approach to containment has been established to test people suspected of exposure to COVID-19. At the same time and separately, the RCGP RSC’s surveillance has been extended to monitor the temporal and geographical distribution of COVID-19 infection in the community as well as assess the effectiveness of the containment strategy.

Objectives

The aims of this study are to surveil COVID-19 in both asymptomatic populations and ambulatory cases with respiratory infections, ascertain both the rate and pattern of COVID-19 spread, and assess the effectiveness of the containment policy.

Methods

The RCGP RSC, a network of over 500 general practices in England, extract pseudonymized data weekly. This extended surveillance comprises of five components: (1) Recording in medical records of anyone suspected to have or who has been exposed to COVID-19. Computerized medical records suppliers have within a week of request created new codes to support this. (2) Extension of current virological surveillance and testing people with influenza-like illness or lower respiratory tract infections (LRTI)—with the caveat that people suspected to have or who have been exposed to COVID-19 should be referred to the national containment pathway and not seen in primary care. (3) Serology sample collection across all age groups. This will be an extra blood sample taken from people who are attending their general practice for a scheduled blood test. The 100 general practices currently undertaking annual influenza virology surveillance will be involved in the extended virological and serological surveillance. (4) Collecting convalescent serum samples. (5) Data curation. We have the opportunity to escalate the data extraction to twice weekly if needed. Swabs and sera will be analyzed in PHE reference laboratories.

Results

General practice clinical system providers have introduced an emergency new set of clinical codes to support COVID-19 surveillance. Additionally, practices participating in current virology surveillance are now taking samples for COVID-19 surveillance from low-risk patients presenting with LRTIs. Within the first 2 weeks of setup of this surveillance, we have identified 3 cases: 1 through the new coding system, the other 2 through the extended virology sampling.

Conclusions

We have rapidly converted the established national RCGP RSC influenza surveillance system into one that can test the effectiveness of the COVID-19 containment policy. The extended surveillance has already seen the use of new codes with 3 cases reported. Rapid sharing of this protocol should enable scientific critique and shared learning.

International Registered Report Identifier (IRRID)

DERR1-10.2196/18606