Edward Orwin Burn

ANTECEDENTS: Sistemes de salut basats en una forta atenció primària organitzada en base poblacional, milloren l’accés als serveis sanitaris, disminueixen el número d’ingressos hospitalaris evitables i per tant la despesa sanitària global, i milloren la salut de la població. L’any 2017, la Comissió Europea va identificar l’avaluació del rendiment dels sistemes d’atenció primària com una prioritat política. mesurar la qualitat dels serveis oferts és un requisit fonamental per a identificar àrees de millora. L’avaluació del rendiment d’un sistema és especialment necessària quan es produeixen canvis substancials en l’estructura o funcionament, com l’ocorregut amb la pandèmia de la COVID-19. A causa de la complexa naturalesa dels sistemes d’atenció primària, els indicadors utilitzats per avaluar el seu funcionament solen ser específics de cada context. Els Ambulatory Care Sensitive Conditions (ACSC) son un conjunt de diagnòstics potencialment evitables d’ingrés hospitalari o mort. Sovint és utilitzat com un indicador de mesura de l’efectivitat de l’atenció primària.
JUSTIFICACIÓ: L’avantatge d’analitzar agrupacions d’EAP en lloc de l’estudi d’aquests EAP a nivell individual és que permet una perspectiva més àmplia i generalitzable de les característiques estudiades.
L’ús d’algorismes d’intel·ligència artificial per predir el número d’ingressos hospitalaris per diagnòstic evitable (ACSC) pot ajudar a identificar els EAP amb problemes d’efectivitat, la qual cosa afavoriria una millora en la planificació de recursos sanitaris.
OBJECTIUS: 1) Caracteritzar grups similars d’EAP de Catalunya en funció dels resultats en salut, els recursos disponibles i les característiques poblacionals. 2) Predir el número d’ingressos hospitalaris per diagnòstics evitables (ACSC) dels diferents EAP. 3) Analitzar i comparar els resultats en salut dels equips d’atenció primària en relació als períodes de la COVID-19: pre-pandèmia, pandèmia i post- pandèmia.
PERÍODE D’ESTUDI: 2018 – 2022.
ENTORN I POBLACIÓ DE REFERÈNCIA: L’ICS és el principal proveïdor de serveis sanitaris de Catalunya. Gestiona 279 Equips d’Atenció Primària (EAP) adscrits a 5,8 milions de ciutadans.
POBLACIÓ D’ESTUDI: EAP dels centres del ICS de Catalunya.
OUTCOMES: Les variables d’interès son les relatives als resultats dels EAP: 1. Accessibilitat; 2. Longitudinalitat; 3. Abast (comprehensiveness); 4. Coordinació; 5. Orientació comunitària; 6. Orientació familiar; 7. Satisfacció; 8. Gestió de la demanda d’infermeria; 9. Benestar emocional; 10. Qualitat de prescripció farmacèutica; 11. Efectivitat: Número total d’ingressos hospitalaris per diagnòstic evitable; Mortalitat i mortalitat per ACSC.
VARIABLES: Referides a les característiques poblacionals : Edat, Sexe, Indicadors socioeconòmics, Número de visites als equips d’Atenció Primària per tipus de professional i EAP, Número total de malalties, Fragilitat (eFi, Efragicap, Nombre de medicaments (segons ATC) dispensats i data, Variables estils de vida, Nombre d’ingressos hospitalaris totals per cada EAP; Referides als recursos disponibles: Recursos humans disponibles de cada EAP per categoria professional i mes, Infraestructura de les EAP: Edificis i pisos, o en el seu defecte, variables que recullin informació sobre la fragmentació del personal de cada EAP, Població assignada.
FONTS D’INFORMACIÓ: El Sistema d’Informació per al desenvolupament de la Recerca en Atenció Primària (SIDIAP) disposa d’informació agregada a nivell mensual i per EAP de les característiques de la població, dels recursos disponibles dels EAP i dels seus resultats en salut.
ANÀLISI ESTADÍSTICA: 1) Per a la caracterització de grups similars d’EAP en funció dels resultats en salut, els recursos disponibles i les característiques poblacionals s’usaran tècniques de clústers. Els clústers resultants es compararan en funció de la distribució de les variables sociodemogràfiques, clíniques, socioeconòmiques, d’estil de vida, i, principalment, d’utilització d’assistència sanitària primària, utilitzant anàlisi de variància (ANOVA) i proves de ji-quadrat. 2) S’elaboraran un conjunt d’algorismes d’intel·ligència artificial per predir el número d’ingressos hospitalaris per diagnòstics evitables (ACSC) dels diferents EAP i per períodes de 1,3 i 6 mesos en períodes de 1,3 i 6 mesos. S’aplicaran les següents tècniques de regressió: Decision Tree (DT), Random Forest (RF), Gradient Boosting (GB), k-nearest neighbors (K-NN), Support Vector Machine (SVM), Artificial Neural Network (ANN) i Recurrent Neural Network (RNN). 3) Per a analitzar i comparar els resultats en salut dels equips d’atenció primària en els tres períodes de temps que comprenen la COVID-19, es realitzaran models ARIMA, suavitzat temporal i ANOVA.
APLICABILITAT I RELLEVÀNCIA: 1) L’agrupació d’EAP pot ajudar a la identificació de patrons de recursos dels EAP que poguessin resultar més efectius per a aconseguir satisfer les necessitats de la població. 2) L’avaluació d’algorismes que realitzin prediccions de resultats en salut permet identificar els EAP que puguin tenir problemes d’efectivitat, la qual cosa afavoreix una millora en la planificació de recursos sanitaris. 3) El tercer estudi podria ajudar a comprendre com la pandèmia ha afectat els resultats en salut durant el seguiment d’aquesta, i si aquests segueixen la mateixa tendència a mesura que es redueix la focalització de l’atenció de problemes relacionats amb la COVID-19 (període post-pandèmia). En global, el present projecte pot ajudar als responsables polítics i als gestors sanitaris a prendre decisions més informades sobre l’assignació de recursos, millorant així la qualitat d’assistència dels EAP i, en conseqüència, la salut de la població.

Objectives: The main aim of this project is to estimate time trends in prevalence and incidence rates, and short- and long-term survival of site-specific cancers in the OHDSI network.
Design: This study will be a multinational observational cohort study and will be conducted using a network of large real world data sources that have been mapped to the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM).
Setting: Population-based, electronic health records, claims and registry data from primary and secondary care.
Participants: Individuals with no prior history of cancer (for incidence and survival analyses only), and who have been on the database for at least 1 year before study entry.
Outcomes: Prevalent and incident cancer diagnoses and overall as well as 1-, 5-, and 10-year survival of site-specific cancers.
Data analyses: The OHDSI Cohort Diagnostics package will be used to assess the fitness of use of cancer data on each database. We will calculate prevalence (PR) and incidence rates (IR) with 95% confidence intervals (95%CI) for each year and study period by dividing the number of ever and first recorded cases of cancer, respectively, by 1,000 person-years of follow-up, overall and stratified by demographics and relevant comorbidities. The overall and 1-, 5-, and -10-year survival rates will be calculated as the percentage of people who have been diagnosed with cancer and are still alive during the study period as well as one or five years after diagnosis, respectively, per year and stratified by pre-defined subgroups. To assess the incidence trend over time, we will calculate the IRs in 5 year periods and then calculate the incidence rate ratios (IRRs) and their corresponding 95%CI to analyze the differences in incidence between the defined time periods.

Rationale and Background
The research agenda of the Vaccine Monitoring Platform jointly coordinated by EMA and the European Centre for Disease Prevention and Control (ECDC) includes the continuous assessment of COVID-19 vaccine effectiveness.
COVID-19 vaccines were authorised for use in the European Union. These vaccines, and any (future) adapted vaccines, would therefore benefit from post-authorisation studies to provide real-world evidence to guide regulatory and vaccination policies. A recent post-authorisation study performed in the Nordic countries, where near real-time data is available, showed that receipt of a bivalent BA4/5 mRNA booster as a fourth dose provides 67.8% protection against COVID-19 related hospitalisation. There is also evidence that effectiveness starts to wane after a few months. For regulatory purposes, such data are especially useful for the most recent variants, including XBB and later.
There is mounting evidence on post-acute outcomes of SARS-CoV-2 infection. This can include very specific outcomes such as cardiovascular events or the incidence of new-onset diabetes, or broader definitions such as the WHO clinical case definition for post COVID-19 condition. Data are needed regarding the COVID-19 vaccines effectiveness at preventing these outcomes. This is pertinent for the most recent variants, but equally important for older variants.

Objective(s)
To generate additional evidence on the effectiveness of COVID-19 vaccines at preventing severe COVID-19 and post-acute outcomes of SARS-CoV-2 infection.

Specifically, this study has 6 objectives:
1. To assess the effectiveness of COVID-19 vaccination for the prevention of severe COVID-19 related outcomes (COVID-19 related hospitalisation or COVID-19 related death)
2. To assess waning of the effectiveness of COVID-19 vaccination for the prevention of severe COVID-19 related outcomes (COVID-19 related hospitalisation or COVID-19 related death)
3. To assess the effectiveness of COVID-19 vaccination for the prevention of all-cause mortality in the 3- and 6-months following discharge for COVID-19 related hospitalisation
4. To assess the effectiveness of COVID-19 vaccination for the prevention of new-onset type 1 Diabetes Mellitus in the 12 months after a SARS-CoV-2 infection
5. To assess the effectiveness of COVID-19 vaccination for the prevention of new-onset type 2 Diabetes Mellitus in the 12 months after a SARS-CoV-2 infection
6. To assess the effectiveness of COVID-19 vaccination for the prevention of cardiovascular events in the 12 months after a SARS-CoV-2 infection

Research Methods
Study design: Population-level cohort studies

Data sources:
• Clinical practice Research Datalink (CPRD) GOLD, United Kingdom
• Integrated Primary Care Information Project (IPCI), The Netherlands
• The Information System for Research in Primary Care (SIDIAP), Spain

Additional databases can be added as part of a routine repetition of this study once they are successfully onboarded for DARWIN EU and meet feasibility requirements for this study.

Exposure:
Covid-19 vaccines BNT162b2 (Comirnaty) and mRNA-1273 (Spikevax), particularly the number of received vaccine doses per brand.
For those databases where this information is available, the 4th dose (2nd booster) will be stratified for monovalent (original strain) or adapted, bivalent (original strain + omicron ba1 or original strain + omicron ba4/5).
Analyses will be conducted separately for each vaccine brand.

Primary outcomes of interest:
1) Outcomes assessed from start of rollout of 4th vaccine dose/ 2nd booster dose program onwards:
1. COVID-19 related hospitalisation
2. COVID-19 related death
2) Outcome/s during periods with dominance of any SARS-CoV-2 variants:
3. All-cause mortality in the 3 months after discharge from a COVID-19 hospitalisation
4. All-cause mortality in the 6 months after discharge from a COVID-19 hospitalisation
5. Incidence of new-onset type 1 Diabetes Mellitus beyond the first 30d after SARS-CoV-2 infection
6. Incidence of new-onset type 2 Diabetes Mellitus beyond the first 30d after SARS-CoV-2 infection
7. Incidence of cardiovascular events (cerebrovascular disorders, dysrhythmias, ischemic and non-ischemic heart disease, pericarditis, myocarditis, heart failure and thromboembolic disease) in the 12 months after a SARS-CoV-2 infection
COVID-19 related hospitalisation (outcome 1, part of outcomes 3 and 4) is not available for IPCI and CPRD but will only be assessed in SIDIAP.

Study population:
All subjects aged 12 years and older, with at least 365 days of data availability before index date (ID) [ID defined as the date of the latest vaccine dose administered] AND data availability from 12/2020 onwards (i.e. the time when the roll-out of the vaccination campaign started) in the respective database will be included.

All studies will be carried out comparing 8 cohorts, which we defined based on varying degrees of vaccine exposure and period of predominant SARS-CoV-2 variant.

Outcomes 1-2 will be assessed in cohorts 1-2 (and 3-4 where available), outcomes 3-7 will be assessed in cohorts 5-8.
Unvaccinated groups will not be used as a comparator in our study for vaccine effectiveness research because they may be very different from vaccinated individuals regarding their risk of infection with SARS-Cov-2. This study therefore focusses on the association of varying degrees of vaccine exposure and COVID-19 related outcomes.

Study period:
Period of “start of roll-out 4th dose/2nd booster dose onwards”: from 01/08/2022 – last available data for each database.
Period of “XBB variant or later dominant”: 01/03/2023 – last available data for each database. Note: This period is not covered by any of the data cuts onboarded for DARWIN EU at the time of protocol submission.
Period of “any variant dominant”: 01/01/2021 (when the wider roll-out of the vaccination campaign started) – last available data for each database.

Statistical analyses:
All analyses will be conducted separately for each database, and will be carried out in a federated manner, allowing analyses to be run locally without sharing patient-level data. For each analysis, we will subsequently pool effect estimates across databases using random effect meta-analyses, I^2 for heterogeneity will be reported.
Cell counts <5 will be suppressed to comply with the database’s privacy protection regulations.

Rationale and Background:

The extended mandate of EMA reinforcing the role of the Agency in crisis preparedness and management of medicinal products and medical devices became applicable on 1st March 2022 (Regulation on EMA’s extended mandate becomes applicable | European Medicines Agency (europa.eu)).
EMA is now responsible for monitoring medicine shortages that might lead to a crisis situation, as well as reporting shortages of critical medicines during public health emergencies (PHE). Such shortages would make it difficult or impossible to meet the treatment needs of individual patients or populations. The Agency has also the mandate to coordinate responses of EU / EEA countries to shortages of critical medical devices and in-vitro diagnostics in crisis situations.
Scientific and commercial data on monthly prescriptions of medicines that may be critical in PHE can help understanding trends and seasonal variations. In conjunction with time series and forecasting models, as well as data on medicines supply, such data will contribute to the on-going efforts of the Agency to better monitor and coordinate its response to shortages of critical medicines.
This study aims at generating monthly prescription rates of selected medicines over the last 10 years and to fit Autoregressive Integrated Moving Average (ARIMA) prediction models to such data.

Research question and objectives:

This study aims to characterise the incidence of use (prescription or dispensation) of 11 antibiotics used for public health emergencies that are considered at risk of shortages in order to understand trends, cycles and seasonality in the use of those medicines; and to forecast short-term prescription rates of such medicines under assumed scenarios, which could help anticipate and prevent potential shortages, or manage them.
The general research question is: What are the monthly prescription rates of selected medicines of importance for public health emergencies over the last 10 years?
The specific objectives of this study are:
(i) To estimate monthly incidence rates of use (prescription or dispensation) of the 11 selected medicines during a 10-year period from the most recent data available, stratified by age and sex, in each of the databases.
(ii) To conduct time series modelling by fitting an ARIMA model to data generated in objective 1 for short-term (6-month) forecasting.

Research Methods:

Study design
• Population level cohort study (Objective 1, Population-level drug utilisation study on antibiotics)
• Population level cohort study (Objective 2, Time series modelling based on the Population-level drug utilisation study on antibiotics)
Population
Population-level drug utilisation of antibiotics: All individuals present in the database in the last 10 years of available data will be included in the analysis after 30 days of database history. For this population, incidence of use of antibiotics will be explored.

Variables
Drugs of interest: list of 11 antibiotics that may be critical in PHE
Calendar month, age, and sex will be used for stratification.

Data sources
1) IQVIA LPD Belgium (Primary Care Database)
2) CPRD Gold (UK, Primary Care Database)
3) SIDIAP (Spain, Primary Care Database)
4) IMASIS (Spain, Secondary Care Database)
5) IQVIA DA (Germany, combination of primary and secondary care (outpatient visits) database).

Sample size
No sample size has been calculated. Based on a preliminary study feasibility assessment the expected number of prescriptions in the period investigated is expected to be between <1Kand 25M across the five data sources considered. Data analyses Population-level drug utilisation study on antibiotics: monthly incidence rates of antibiotic use per 100,000 person-year, as described in section 8.7.5.1 – Population-level drug utilisation study. Time series modelling: forecast of the 6-month incidence rates of antibiotic use after the end of available data in the data source using AutoRegressive Integrated Moving Average (ARIMA) models for time series analysis, as described in section 8.7.5.2.

Serious adverse events have been reported among antidementia drug users. We aim to analyse the use and comparative safety of the antidementia drugs in primary care centres in Catalonia (Spain).
Design: A population-based cohort study using real-world primary health care data (SIDIAP database), standardized to the Common Data Model OMOP with hospital linkage (CMBD database). A nested case-control approach will be adopted to assess risk of adverse events (AE) during the study period (2007-2020). Inclusion criteria: for the drug utilization study (DUS) we will include individuals with at least 40 year or older with dementia, registered for at least 1 year before cohort entry and with at least 1 prescription of rivastigmine, galantamine, donepezil or memantine during study period. For the safety study we will include patients from the DUS with an incident use (no previous 365 days use) of any of the study drugs. For the nested case control analysis, we will create a cohort of patients with dementia without treatment (anytime). Exposure (DUS/Safety): prescription of rivastigmine, donepezil, galantamine or memantine. Outcomes: (DUS) demographics, comorbidities, prescriber-type, prescribing pattern and proportion of ‘prescription cascade’ drugs (prescriptions generated to alleviate adverse events). Outcomes (safety study): AEs related to disorders of the skin, cardiovascular, gastrointestinal, neurological, psychiatric, sleep, urinary and respiratory disorders, falls, hospitalizations and all-cause mortality.
Statistics: Yearly age-sex incidence rates (IR) and Kaplan Meier curves to assess the duration and drug discontinuation. Adjusted and unadjusted IR of AEs, Hazard Ratios (95% confidence intervals (CI)) using Cox proportional hazard models, and Odds Ratios (95% CI) using nested case-control analysis with conditional logistic regression will be assessed. Each case will be matched (age, sex, Charlson Comorbidity Index and socioeconomic status) with up to 10 controls.
Expected results: With this large population-based cohort study based on routinely collected primary care data with hospital linkage we expect to provide an in-depth characterization of the population that uses any of the currently commercialised antidementia drugs. Because we are using routinely collected real-world information, which differs from the information gathered in randomized controlled trials, we will be able to provide a more accurate picture of the patients that finally receive the drugs to treat dementia. We will also analyse the proportion of patients with a ‘prescription cascade’ and this will contribute to the identification of adverse events that have been misclassified as new medical conditions. Finally, the comparative safety study carried out will inform us about the risk of adverse events between users and non-users of antidementia drugs. Given the increasing availability of non-pharmacological treatments to treat dementia, this information will help clinicians to assess the risk-benefit of these drugs. The other comparative safety studies, between acetylcholinesterase inhibitors versus N-methyl-D-aspartate receptor antagonists and between galantamine and donepezil versus rivastigmine will also contribute to decision making of clinicians and can be the base of other prospective head-to-head comparison studies.
Relevance and aplicability: Population-based cohort studies that use routinely collected health care data have gained importance in the last years because of their ability to provide evidence that could not be generated through randomized controlled trials. This is more important in pharmacoepidemiology given that subjects that are included in the pivotal trials of the drugs usually differ greatly from the overall population that finally receives the medication. The ageing of the populations foresees an increase in the cognitive related disorders including dementia and Alzheimer disease. Although the actual medications to treat dementia (rivastigmine, galantamine, donepezil and memantine) were commercialized more than a decade ago, few are the studies that have analysed their real-world adverse effects at a population level, none of them in Spain, and even fewer are the ones that have done this comparing the different medications. Regardless of the new therapies that are being recently developed to treat dementia the great burden of the treatment still relies on these four medications and therefore is it necessary to fully identify, describe in depth the patients that are using these drugs. The comparative safety analysis results will help health care providers and clinicians to assess the risk-benefits of these drugs to avoid future adverse events or better target patients who receive medication. At last, this can also be a hypothesis generating study given that it will help identify the characteristics of the population at an increased risk of adverse events which could help design future antidementia drug studies.

We aim to estimate the risks of long-term COVID-19 outcomes among individuals with COVID-19 or exposed to COVID-19 during pregnancy, as well as the effect of vaccines on the development of long COVID and long-term COVID-19 outcomes.
Data will be obtained from SIDIAP, which includes primary care records for approximately 6 million people in Catalonia. To estimate long-term COVID-19 outcomes (Objective 1), we will conduct three population-based matched cohort studies using a target trial emulation design from September 2020 to December 2023. We will match COVID-19 infections to uninfected controls in a 1:5 ratio using propensity score (PS) matching. Cases/controls matched cohorts will include: Objective 1.1: people not vaccinated against COVID-19; Objective 1.2: people vaccinated against COVID-19; and Objective 1.3: newborns (cases: exposed to COVID-19 infection during pregnancy, controls: unexposed). Cohorts will be followed for 2 years. Our outcomes will include autoimmune, cardiovascular, mental, neurological, renal and early-life complications. We will estimate cause-specific Hazard Ratios (HR) for each outcome.To estimate vaccine effectiveness on the development of long COVID, we will first characterise long COVID based on persistent symptoms for >28 days (Objective 2.1). We will then estimate the effect of vaccines on the development of long COVID (Objective 2.2) using a staggered cohort study design. Vaccinated and unvaccinated cohorts will be compared using different PS techniques. We will then calculate HR for long COVID and long-term outcomes.
Our findings will inform preventive strategies and post-acute care pathways, thus contribute to prevent long COVID and improve COVID-19 survivors’ health.

The global pandemic of COVID-19 has resulted in over 50 million reported cases and over 1.2 million deaths globally. Meanwhile, hundreds of clinical trials of vaccines are ongoing and some of them show clinical efficacy. While planning for the large-scale immunization program, it is important to understand the potential adverse events after vaccination or viral infections. Electronic health records have been increasingly used in safety study, including SIDIAP. The ability to and the reliability of capturing the adverse events using suitable phenotyping algorithms in such databases is the foundation in conducting these studies. We will firstly identify the phenotyping algorithms of the AESI used in other studies, or develop the phenotypes if no existing one is found. Then we will evaluate the performance of these phenotypes using the diagnostic and evaluation tool that had been previously developed. The second objective is to estimate the background incidence rates (IR) of the AESI among the general population from year 2006 to 2019. Individuals who were observed for at least 365 days in the dataset during the study period will be included. The numerator of the incidence rate will be the total number of incident cases in a given year, and the denominator will be person-time at risk in each year. We will also estimate the IR among patients who were diagnosed or received a positive test for COVID-19 (after February 2020) or seasonal influenza. We will apply different algorithms in identifying both the exposures and the outcomes. We will also conduct a self-control case series analysis to explore the association between developing the AESI and COVID-19 or influenza infections.

Background: The overwhelming numbers of COVID-19 infections have led to an unprecedented crisis for healthcare systems worldwide. Understanding the impact of this pandemic on health is an urgent priority.
Objectives: UNCOVER aims to unravel the mid- and long-term effects of the pandemic on COVID-19 and non-COVID-19 morbidity and mortality through the use of large real-world databases from Catalonia and the UK.
Methods: UNCOVER will conduct a cohort study using longitudinal electronic health records from Catalonia (SIDIAP) and the UK (CPRD). These databases have been transformed to an international common data model (CDM), and when possible, the study will be replicated in other databases around the world. COVID-19 testing, diagnosis in primary care, hospitalisations, and deaths will be identified from March 1st 2020. Study periods will be defined as pre-COVID-19, COVID-19-pre-vaccine, and COVID-19-vaccination. Different periods will be defined for the study of non-pharmaceutical interventions by country. Mid- and long-term symptoms and outcomes will be identified. Time-series and multi-state models will be used for data analyses.
Expected results: UNCOVER will provide novel scientific knowledge aimed at helping decision-makers and clinicians in the control and management of the pandemic on a national and international level, whilst improving populations’ health and quality of life.

Background
To further monitor COVID-19 vaccine safety and complement pharmacovigilance measures, multi-national observational studies have been requested by the EMA: Incidences of patient-reported side effects after COVID-19 vaccination and adverse events of special interest are closely being monitored. The Covid-Vaccine-Monitor project will facilitate the rapid signal assessment of emerging safety concerns.

Hypothesis
These existing initiatives will provide important data on the incidence of adverse outcomes reported after vaccination and on potential risk factors for thromboembolic events in COVID-19 patients.

Objectives:
1-a)To quantify the association between the administration of a COVID-19 vaccine and the occurrence of thrombosis with thrombocytopenia syndrome/s (TTS) within pre-specified risk periods, stratified by vaccine type/brand, age and gender, while controlling for relevant confounding factors.
1b) To quantify the association between different COVID-19 vaccine types/brand (where possible/applicable), while controlling for relevant confounding factors.
2a) To quantify the association between the administration of a COVID-19 vaccine and the occurrence of thromboembolic events (TE) within pre-specified risk periods, stratified by vaccine type/brand, age and gender, while controlling for relevant confounding factors.
2b) To quantify the association between different COVID-19 vaccine types/brands (where possible/applicable), while controlling for relevant confounding factors.
3) To study the association between pre-specified potential risk factors and TTS in people receiving COVID-19 vaccine/s
4) To characterize the treatments used in patients with TTS, including the use of anticoagulants and other therapeutic products
Exploratory objective:
5) To develop a proof-of-concept study to support future genetic and pharmacogenomic analyses of the association between COVID-19 vaccines and thromboembolic events and TTS. The three specific subobjectives are:
Objective 5.1. To identify genetic variations associated with TE based on previous literature and a review of previous GWAS studies
Objective 5.2. To investigate whether the risk of TE following COVID vaccination is modified by genetic susceptibility
Objective 5.3. To assess the feasibility of identifying TTS in UK Biobank using linked hospital and primary care records to inform future pharmacogenetic studies

Methods:
Data will be obtained from five European primary care records, two outpatient records, and one inpatient records databases (UK,Spain,The Netherlands, France, Denmak). In addition, one US claims and one large US hospital records database will be accessed to maximise sample size and exposure to vaccines currently under-represented in European data.All of the data sources are previously mapped to the Observational Medical Outcomes Partnership (OMOP) common data model (CDM).

Distributed network cohort studies will be conducted to answer objectives 1-4. Propensity score matching based on large-scale propensity scores will be used to minimise confounding by indication for objectives 1-2 and 5

The study period will cover from Dec 2020 (first vaccine users) until the latest data extraction available in each of the contributing databases. For objectives 1-3, lags in hospital linkage will result in two different study periods for analyses based on primary care vs linked data in both CPRD AURUM, GOLD and SIDIAP.
For other objectives and data sources the study period will be unique and will go from cohort-specific index date to the latest data available.
Cohort-specific index dates are: For vaccinated people (and matched unvaccinated) [Objectives 1a-2a, 3, 5.2 and 5.3]: date of first dose vaccine (and same date for matched unvaccinated);For comparative cohorts [Objective 1b and 2b]: date of first dose of the corresponding vaccine type/Brand; For TTS cohorts [Objective 4]: date of TTS diagnosis.

Source and study population:
Target population: All persons registered in any of the contributing databases within the study period and with at least one year of data visibility before December 2020 will be eligible.
Study population for Objectives 1a-2a, 5.2 and5.3: Of these, people with at least one exposure to any COVID vaccine in the study period will be included in the ‘exposed’ cohort/s, with 1st and 2nd dose vaccine date as time-varying index dates. Unexposed matched groups will be pooled from the Target population.
Study populations for Objective 1b & 2b: Those with at least one exposure to viral vector-based vaccines will be included in the exposed group/s and those with at least one exposure to mRNA COVID-19 vaccines in the active comparator group/s. Similarly, those with Vaxzevria vaccine will be included in the exposed group and those with Comirnaty as the active comparator in the vaccine brand comparative safety analyses based on CPRD AURUM, GOLD, RCGP, and SIDIAP.
Study population for Objective 3: Those with at least one exposure to any COVID vaccine in the study period will be included in this cohort for the analysis of risk factors of post-vaccine TTS.
Study population for Objectives 4: Those with a TTS event in the up to 28 days post vaccination of any dose will be included as ‘TTS cases’ for Objective 4, with TTS date as index date.
Study population for Objective 5: UKBB participants with linkage to primary care and HES data and vaccine exposure identified in the primary care records will be included in the exposed group. Unexposed matched participants will be pooled from those with linked primary care and HES data

Data analysis
All the analyses detailed below will be conducted stratified by database and by age, gender and vaccine type/brand

The use of healthcare data, generated through the delivery of normal clinical care is increasingly being proposed as a source of evidence to support not only drug development and regulatory decision-making but also to understand the physiology and pathogenesis of diseases.
Use of multiple electronic health care databases is important not only to increase sample size but also to investigate country specific differences, differences by type of databases (e.g. primary vs. secondary care) or to replicate findings. One of the challenges however are the differences between the databases with regard to the underlying structures and semantic mapping. A common data model could help harmonise healthcare data across multiple data sets and provide a mechanism to allow the conduct of multi-database, international studies.
The European Health Data and Evidence Network (EHDEN) project (https://www.ehden.eu/) is an international project supported by the Innovative Medicines Initiative (IMI) aiming to standardize health care data to the Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) and to develop and implement tools to facilitate research on large electronic health care databases.
One of the objectives of the EHDEN project is to test existing methodologies but also to develop new methodologies and analytical tools to conduct (pharmaco)epidemiological research using electronic health care databases mapped to the OMOP CDM. To investigate the validity and functionality of this approach, we want to conduct a drug-utilisation study using EHR data. As proof of concept study we want to conduct a drug utilisation studies on respiratory drug use in patients with asthma and chronic obstructive pulmonary disease (COPD). This research is important and relevant as asthma and COPD are prevalent conditions, primarily treated in primary care.

Objectives: First, to describe the features and characteristics of the complications of viral infections, with a particular focus on viral pneumoniae. Second, to assess the predictors of adverse outcomes amongst patients with virus-related hospitalization and generate algorithms to identify subjects most at risk of complications and/or morbi-mortality. Third, to compare the safety of treatments being considered/used for potential use in COVID-19 (including hydroxychloroquine, systemic steroids, and ACE inhibitors (angiotensin converting enzyme inhibitors)/ ARBs (angiotensin-receptor blockers)).
Design: This study will use SIDIAP data linked to ICS CMBD hospital data, and mapped to the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). Study cohorts will be generated, descriptive statistics will be used for the first research question, prediction modelling will be used for the second research question, and a propensity-score adjusted analysis will be used in the third question.
Setting: Population-based, electronic health records from primary and secondary care in Catalonia.
Participants: Individuals diagnosed with a viral infection.
Exposure: For research questions 1 and 2, exposures will be a record of a viral infection. For research question 3, exposures will be incident use of a medication of interest.
Outcomes: Viral infection, viral pneumonia, adult distress respiratory syndrome, advanced life support including ICU and/or mechanical ventilation or intubation, all-cause death.
Applicability: The results of this study will have an immediate impact on the response of COVID19.
Limitations: Selection bias could affect the first two analyses and confounding by indication could affect the third. Appropriate analytical methods will be employed to minimize such issues.

Background
Conventional synthetic disease modifying anti-rheumatic drugs (csDMARDs) are usually the first line of treatment of rheumatoid arthritis (RA). This study has the following research objectives: 1) to characterise treatment patterns in rheumatoid arthritis, 2) to predict the risk of safety outcomes for individuals initiating treatment with csDMARDs, and 3) to assess the comparative safety of alternative first-line csDMARD treatment strategies commonly used in rheumatoid arthritis.

Methods
Study participants will have a drug utilisation record of csDMARD after a diagnosis of RA, with the first such record taken as their index event.
First, individuals’ treatment pathway involving csDMARDs after a diagnosis of rheumatoid arthritis will be characterised. Second, prediction models will be developed for the 5-year risk for safety outcomes of interest. A range of algorithms will be used to develop these prediction models (for example, regularised logistic regression and Random Forest). Lastly, the relative safety of common first-line csDMARD treatment strategies will be compared. Cox models will be estimated, with propensity score adjustment used to control for confounding by indication.

Relevance and limitations
The results from this study will help to improve our understanding of current prescribing practice in RA and help to inform shared decision making. However, a limitation of the study on treatment pathways will be that biological DMARDs are not expected to be observable in SIDIAP. Meanwhile, with factors such as radiographic evidence also unavailable in SIDIAP, it will not be possible to include these unobserved factors in the generation of propensity scores.

El projecte inclou dos estudis:
El primer estudi, ja fet fa una setmana durant 4 dies amb diverses bases de dades internacionals online, ha estudiat el comportament d’infeccions víriques similars que hi ha hagut en el passat: s’han descrit les característiques de les persones amb complicacions d’infeccions víriques com la grip, s’han valorat els predictors de resultats adversos entre els pacients hospitalitzats amb pneumònies virals, s’han generat algoritmes per identificar els pacients amb més risc de complicacions i/o morbimortalitat, i s’ha avaluat la seguretat dels tractaments utilitzats per a un ús potencial en Covid-19.
El segon estudi vol descriure les característiques de les persones amb Covid-19 a Catalunya, així com desenvolupar models predictius de les complicacions de la Covid-19 fent servir els mètodes i resultats obtinguts en el primer estudi. Per això, es farà la integració de la informació dels pacients infectats amb Covid-19 a Catalunya, la transformació de la nova informació a un model de dades internacional, i després de contrastar les dades amb altres bases de dades d’altres països, s’elaboraran uns criteris pronòstic aplicables a les polítiques de control de pandèmia de Covid-19 al nostre país.
Els models predictius de casos greus d’infecció vírica permetran classificar els pacients amb Covid-19 per gestionar-los adequadament, avaluant la necessitat de que el malalt vagi a l’hospital o es quedi a casa, i les condicions necessàries per al seu tractament.

A goal of Discovery and Translational Sciences is to implement new technology platforms to accelerate research. This grant allows the rapid acquisition and analysis of emerging data from the ongoing global outbreak of Covid-19. The clinical and epidemiological data will inform the foundation’s response to the outbreak including expanding our understanding of risk factors for disease progression and the design of efficient clinical trials.

Objectives: To investigate the association between bariatric surgery (BS) and the risk of obesity-related cancers (esophagus, liver, pancreas, colorectal, breast (postmenopausal), endometrium, kidney, stomach, gallbladder, ovary, thyroid, meningioma, and multiple myeloma) and to examine the association between BS with all-cause mortality.

Design: This study will be a matched cohort using SIDIAP and CMBD-PADRIS source data and data been mapped to the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM).

Setting: Population-based, electronic health records from primary and secondary care in Catalonia.

Participants: Individuals aged ?18 years, with a record of BMI ?35 kg/m2 or an Obesity diagnosis, without history of bariatric surgery nor cancer, who have been on the database for 1 year before study entry and have at least 1 year of follow-up.

Exposure: Bariatric surgery procedure.

Outcomes: Incident diagnoses of obesity-related cancers and date of death.

Expected Results: Participants who undergo BS have a decreased risk of obesity-related cancers and improved survival compared to morbidly obese participants who do not undergo bariatric surgery.

Applicability: The results of this study will be valuable for policymakers (decision-making about financing this procedure), for surgeons (surgery recommendation), for patients (knowing the risks and benefits of this procedure) and for general practitioners (advising patients to consult a surgeon).

Limitations: Insufficient statistical power to study less frequent cancer types or to find suitable matches of cases and possible unmeasured confounding due to variables that might be lacking in the SIDIAP database.

We aim to develop a prediction model to assess individuals’ risk of all-cause mortality after SAVR or TAVR and, to assess the effectiveness and safety of TAVR vs SAVR emulating the PARTNER 1A trial conditions using routinely collected data. For this, we will use the OMOP maped SIDIAP database to perform a cohort study including all patients that had undergone a SAVR or TAVR intervention between 01/01/2013 and 01/09/2019 and would have been eligible for the PARTNER 1A trial (high mortality risk patients). Our main outcome will be 1-y mortality and we will look at the secondary clinical outcomes defined by the Valve Academic Research Consortium (VARC-2).
For our first aim, we will train a range of algorithms and perform internal and external validation of the results. For our second aim we will use a Cox model controlling confounding using propensity score methods (IPW, matching and stratification). We will also assess if there is heterogeneity of treatment present. We expect to be able to develop an accurate short-term mortality prediction model. We expect to find similar 1y mortality and lower safety outcomes rates in the TAVR group compared to the SAVR group. Our study could be limited by the lack of clinical variables like ejection fraction or presence of bicuspid valve, that could lead to confounding. This will help inform clinical practice and decision makers about safety risks and effectiveness of interventions for valve replacement.

Background
Clinical trials have shown an association between aromatase inhibitor (AI) use for the treatment of breast cancer and adverse musculoskeletal disease when compared with tamoxifen. This has not been investigated in routine clinical practice.

Hypothesis
In comparison to tamoxifen, AI use is associated with an increased incidence of carpal tunnel syndrome (CTS), tendinopathy, osteoarthritis and related procedures/surgery in post menopausal women with hormone receptor positive breast cancer.

Methodology
Cohort study design, using non identifiable SIDIAP OHDSI CDM mapped data in order to enable replication of the study internationally within the OHDSI network. All post-menopausal (defined >55 years) women with an incident diagnosis of breast cancer, with 6 months continous enrolment prior to diagnosis, and incident use of AI, tamoxifen, or both within 1 year of diagnosis. Those with a prevalent cancer, or outcome prior to breast cancer diagnosis are excluded. Incidence of fracture (vertebral and appendicular fractures) to be used as a positive control outcome.

Variables & Measurements
The date of exposure will be identified from the first drug exposure; outcomes defined as the first incident record of an outcome, each outcome assessed independently, identified using OMOP CDM ATLAS generated definitions.

Statistical Analysis
Propensity score adjustment to minimise confounding. Cumulative Incidence of outcome; cox proportional hazards modelling to estimate hazard ratios for each of the outcomes.

Expected results
Incidence of adverse musculoskeletal outcomes at a population level in Catalonia, with replication worldwide to enable international comparison.

Applicability
To counsel women at the beginning of treatment; consider earlier identification and surveillance during treatment.

Strengths & Limitations
The study is based upon drug dispensation rather than adherence, and only contains information from clinical services, leading to the potential to underreport an association. The study aims to increase generalisability through using OMOP CDM mapped data, to ensure the study can be replicated within the international OHDSI community.

Background: Coronavirus disease-2019 (COVID-19) patients appear to be at an increased risk of venous and arterial thromboembolic events. There is a need to better understand the risks of thromboembolic events among patients with COVID-19, their impact on prognosis, the risk factors for such events, and whether individuals’ risks can be predicted given their demographic characteristics and medical history.
Objectives: To estimate the incidence of thromboembolic events among patients with COVID-19, to calculate the risks of worsening of COVID-19 stratified by the occurrence of a thromboembolic event, to assess the impact of risk factors on the rates of thromboembolic events among patients with COVID-19, and to develop patient-level prediction models for venous thromboembolic events for patients with COVID-19.
Design: This study will use SIDIAP data mapped to the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) as part of a European international network cohort study.
Setting: Population-based, electronic health records from primary and secondary care in Catalonia.
Participants: Individuals diagnosed with COVID-19 or tested positive for SARS-CoV-2.
Exposure: A recorded diagnosis of COVID-19 or positive test for SARS-CoV-2.
Outcomes: Venous thromboembolic events, arterial thromboembolic events, cardiovascular events, mortality.
Applicability: The results of this study will have an immediate impact on the management of COVID-19.
Limitations: Ascertainment bias may make any comparisons with estimated rates for historical comparators from the literature difficult.

Osteoarthritis (OA) is the most common arthritis and a major cause of disability in older people. However, apart from a few cross-sectional studies, little research has been done into its comorbidities. A recent meta-analysis study from Weiya Zhang’s team in Nottingham found out that 67% people with OA have other chronic diseases, which is about two times more than an age/gender matched control without OA. However, whether these comorbidities just co-exist with, share common risk factors, cause or are consequences of OA remains unknown. Therefore, we propose a multi-centered research project of cohort and/or case-control studies using primary care data from SIDIAP platform (Information System for the Development of Research in Primary Care). We aim to examine 1) prevalence and incidence of comorbidities in OA and time sequence between OA and comorbidities; 2) common clusters and impact of comorbidities in people with OA; 3) association between commonly used OA analgesics and comorbidities and these three aims will be answered in three working packages (WP1-3). The statistical analysis plan are described in more details in the methodology section V of this protocol.
In this multi-centered project funded by FOREUM (Foundation for Research in Rheumatology), the SIDIAP platform will be used for WP 1-3, together with the other three national/regional registration databases in the UK, Netherlands, and Sweden. Finally, data from different countries will be meta-analysed for consistency among countries (WP5). The heterogeneity and generalisability of the results (as well as other sites) will be a crucial issue and need further discussion.

? Study Title
Estimating Short-Term and Long-Term Direct Economic Burden Associated with Osteoporotic Fractures
? Background and Rationale
Osteoporotic fractures (OFs) among adults are considered an important public health concern, with up to 50% of women and 22% of men over the age of 50 years experiencing at least one fragility fracture in their lifetime. Additionally, the risk of osteoporotic fractures increases with age, especially in postmenopausal women among whom decreased estrogen levels are associated with decreased bone mineral density (BMD). Osteoporotic fractures are associated with significant burden both in formal (e.g. hospitalizations, rehabilitative services, long-term care) and informal (e.g. care provided by family and friends) care settings, and increased mortality. Considering the total burden of OFs to both patient and society, it is important to better understand the short- and long-term direct healthcare impact in order to enhance osteoporosis management in the contemporary care setting. This study will focus on evaluating the direct economic burden of OFs, while a companion protocol will evaluate the indirect and humanistic burden of OFs. Moreover, the outputs from the proposed study will inform policy makers, clinicians, and patients about the multi-national burden of OFs in women, and help payers and clinicians understand the importance of treatment advances that can reduce the risk of osteoporotic fractures. This study will have the advantage of estimating the direct economic burden in six countries using the same study population composition and same time period rather than previous studies that have varying study population characteristics at varying time periods.

? Research Question and Objectives
The study aims to evaluate the direct economic burden of OFs in women aged 50 years or older in the short-term (index fracture date to 12 months) and long-term (one year to five years) following the date of the first recorded osteoporotic fracture.
Study Objectives
1. Estimate the short-term and long-term direct, post-index healthcare resource utilization (HCRU) and costs in women who experienced an incident OF and in a matched cohort of women free of any OF.
2. Estimate the short-term and long-term direct HCRU and costs pre- versus post-fracture (before and after osteoporotic fracture) in women who experienced an incident OF.
? Study Design/Type
This is a multi-national, retrospective cohort study to assess direct economic burden of OF between 2013-2018 in women aged 50 years or older in 6 countries (Australia, France, Germany, Japan, Spain, and USA). The direct all-cause HCRU and cost experience of women with an incident OF in each country will be compared with matched women without an OF (i.e. non-OF cohort) during the study period. Because of differences in healthcare systems, provision of services and costs across countries, outcomes will be reported by individual country. Women with an osteoporotic fracture will initially be matched to women in the non-OF cohort using the same birth month and year as the women with an osteoporotic fracture, then matched 1:1 on selected variables as described in Section 9.1.3. Additionally, among women with an OF, pre-fracture and post-fracture HCRU and costs will be compared. Regional or national electronic medical records (EMR), registries, or claims databases will be used in each country as described in section 10.1. For example, the PharMetrics Plus, anonymised patient-level claims database including primary and secondary care data from US commercial payers, will be used for the USA.
The study design is depicted in Figure 1 above; the following time periods/dates have been defined:
1. Index date: The index date for the OF cohort corresponds to the calendar date of the first record of an OF (i.e. incident OF) between January 1, 2013 and November 30, 2018. The index dates of the women with osteoporotic fracture will then be assigned to the corresponding matched non-OF women, in order to ensure there are no temporal differences in the comparisons between the populations.
2. Pre-index period: The pre-index period (i.e. baseline period) corresponds to the 18 months preceding the index date for both the OF and the non-OF cohort. Women must have 18 months of continuous enrollment in the database pre-index to ascertain osteoporotic fracture-free status and comorbidity and medication history.
3. Follow-up period: The follow-up period corresponds to the period extending from the earliest of index date up to December 31, 2018 (study end date), death, fracture event in non-OF woman, or lost to follow-up (drop out of the database). For woman included in the study, the follow-up period can range from a minimum of one month up to six years from the index date.
? Study Population
The study population is women aged ?50 years with or without an osteoporotic fracture. This study will be conducted in six countries: United States, Australia, France, Germany, Spain and Japan.

? Patient Eligibility
Inclusion Criteria:
OF cohort:
1. Women aged ?50 years when experiencing an incident osteoporotic fracture at the following skeletal sites: hip, vertebral (spine), forearm (radius, ulna), humerus, pelvis, proximal femur, tibia, fibula, ribs, clavicle, scapula, and ankle between January 1, 2013 and November 30, 2018.
2. Continuously enrolled in the database for at least 18 months prior to index date and at least 1 month after index date.
3. No osteoporotic fracture in the pre-index period (i.e. 18 months prior to index fracture)
Non-OF cohort:
1. Women aged ?50 years and with no record of osteoporotic fracture in the pre-index period (i.e. 18 months prior to assigned index date).
2. Continuously enrolled in the database for at least 18 months prior to index date and at least 1 month after assigned index date.
Exclusion Criteria for both cohorts:
1. Record of participation in a clinical trial pertaining to an osteoporotic treatment ?18 months before the index date.
2. Cancer (except non-melanoma skin cancer) during the study period (July 1, 2011-December 31, 2018). Patients with cancer will be excluded because of the high healthcare resource utilization and costs of cancer care as well as effect of cancer and chemotherapeutic agents on bone.
3. Paget’s disease of the bone, osteitis deformans, and osteopathies or metabolic bone diseases (e.g., osteomalacia, hyperparathyroidism, osteogenesis imperfecta) during the study period (July 1, 2011-December 31, 2018). Patients with other bone conditions will be excluded to ensure that the outcomes are associated with osteoporosis and not other bone diseases.

? Matching criteria
Women without OF will be matched to women with OF. Women in the non-OF cohort will first be matched to women with OF using their birth month and year. The index date of the women with fracture (i.e. fracture date) will then be assigned to the corresponding matched non-OF women. After identifying an age-matched group of non-OF patients for each OF patient, the closest matching non-OF woman will be identified through propensity score matching using important confounders (e.g. geographic region, race/ethnicity, total months since index date (fracture date), pre-index glucocorticoid use, pre-index hormone replacement therapy, pre-index anti-osteoporosis drug use, selected comorbidities, and pre-index hospitalizations. The OF and non-OF women will be matched 1:3. If a non-OF woman has a fracture during follow-up, then she will be censored on the date of her fracture. A 1:3 matching will be used to optimize follow-up time of women with OF because a non-OF woman may fracture and be censored before the end of follow-up of the matched woman with OF.
? Variables
Study Outcomes
? Direct all-cause healthcare resource utilization: HCRU will include any resource/services directly provided by the healthcare system in each relevant country, including hospitalisations, emergency room (ER) visits, physician visits, diagnostic and/or reimbursed procedures, and prescriptions. Physical and/or occupational therapy services also will be reported.
? Direct all-cause healthcare costs will be estimated using country-specific costs, and include total direct costs (medical + pharmacy), total medical costs (inpatient + outpatient), hospitalizations, ER, physician, and outpatient pharmacy costs.
? Study Sample Size
The half-length of the 95% confidence intervals (CI) for estimated direct costs was calculated based on mean total costs from the multi-national study of Svedbom et al 2013. The half length of the CI was estimated to be 536 for a sample size of 2,000 and 107 for a sample size of 50,000 for the first year cost of approximately $14,335. The half length of the CI was estimated to be 160 (sample size of 2,000) and 71 (sample size of 50,000) for the fifth year cost of approximately $4,421. The half length for the intervening years since fracture (i.e. 2-4 years) fell between the ranges for the first and fifth year. It is estimated that the number of incident fractures may range from about a low of 16,000 in Australia to more than 140,000 in Spain. Due to the large sample size, the CIs will be narrow, and it is believed that the estimate of direct costs will be with very small estimate error and reliable.
? Data Analysis:
All analyses will be country-specific and will not be combined across countries due to differences in healthcare systems. Demographics, baseline clinical characteristics, and pre-index direct all-cause HCRU and costs will be reported for OF and non-OF groups using number and percent within category for categorical variables, and mean (standard deviation [SD]) with 95% confidence interval or median (interquartile range [IQR]), minimum and maximum values for continuous variables as appropriate. Methods for dealing with missing data such as multiple imputation or last observation carried forward (LOCF) will not be applied, and the number with missing data reported.
To evaluate direct all-cause HCRU and costs among the propensity score matched women who experienced an OF and those who did not, descriptive measures including the mean (SD), median (IQR), and range (minimum, maximum) will be reported. Costs will be log-transformed to diminish the effect of outliers. Direct HCRU and costs will be reported by the year since index date (fracture date) (e.g. ?1 year, >1 to ?2 years, >2 to ?3 years, >3 to ?4 years, >4 to ?5 years since index date) and include all patients alive at the start of each annual period to assess short-term and long-term economic burden of OF. The main outcome is the difference between direct all-cause HCRU and costs in OF and non-OF cohorts (incremental costs). Likewise, the mean HCRU and costs among women with an OF will be compared between 1-year pre-fracture versus each year (1 to 5) post-fracture. Rate of HCRU will be calculated for each year since index date (fracture date) as the number of utilizations divided by follow-up time in the year. The rate will be reported for each individual healthcare resource type. Also, the proportion of women with at least 1 utilization for each resource type (e.g. had at least 1 hospitalization, at least 1 ER visit) will be reported
Comparisons will be made for all osteoporotic fracture types combined as well as by individual osteoporotic fracture types. Differences in HCRU and costs between OF and non-OF cohorts and pre-index versus post-index among OF women will be assessed using regression modelling. A linear regression model with log-transformed costs or gamma regression will be considered. Outcomes will be stratified by residence (i.e. community-dwelling or not) at index date, and also by occurrence of subsequent osteoporotic fracture among OF women during follow-up (yes/no).

The European Health Data and Evidence Network (EHDEN) invites Data partners in Europe to apply for funding to map their health data to the OMOP common data model (CDM). The ambitions of the EHDEN project are high. We aim to standardise more than 100 million patient records across Europe from different geographic areas and different data sources. Mapping of such data to the OMOP CDM will facilitate their use for a variety of purposes, enhancing and accelerating research and healthcare decision-making for global benefit.