Open access Protocol

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
EXAcerbations of COPD and their
OutcomeS on CardioVascular diseases
(EXACOS-­CV) Programme: protocol of
multicountry observational cohort
studies
Clementine Nordon ‍ ‍,1 Kirsty Rhodes,1 Jennifer K Quint ‍ ‍,2
Claus F Vogelmeier,3 Sami O Simons ‍ ‍,4 Nathaniel M Hawkins,5
Jonathan Marshall,1 Mario Ouwens,1 Edeltraut Garbe,6 Hana Müllerová ‍ ‍1

To cite: Nordon C, Rhodes K, ABSTRACT

Quint JK, et al. EXAcerbations Introduction In patients with chronic obstructive STRENGTHS AND LIMITATIONS OF THIS STUDY
of COPD and their OutcomeS pulmonary disease (COPD), the risk of certain ⇒ Retrospective longitudinal cohort studies will be
on CardioVascular diseases run in Germany, Spain, Italy, the Netherlands, the
cardiovascular (CV) events is increased by threefold to
(EXACOS-­CV) Programme: UK, Japan and Canada allowing us to explore the
fivefold in the year following acute exacerbation of COPD
protocol of multicountry
(AECOPD), compared with a non-­exacerbation period. consistency and differences of results across differ-
observational cohort
studies. BMJ Open While the effect of severe AECOPD is well established, ent healthcare systems, database types and patient
2023;13:e070022. doi:10.1136/ the relationship of moderate exacerbation or prior populations.
bmjopen-2022-070022 exacerbation to elevated risk of CV events is less clear. We ⇒ The inclusion of large population-­based cohorts of
will conduct cohort studies in multiple countries to further patients, through the use of electronic medical re-
► Prepublication history and
characterise the association between AECOPD and CV cords and claims databases, will minimise selection
additional supplemental material
events. bias and allow sufficient statistical power to explore
for this paper are available
online. To view these files, Methods and analysis Retrospective longitudinal cohort different categories of cardiovascular (CV) events
please visit the journal online studies will be conducted within routinely collected separately.
([Link] electronic healthcare records or claims databases. The ⇒ Misdiagnoses of respiratory events with CV events
bmjopen-2022-070022). study cohorts will include patients meeting inclusion are likely in the first days following exacerbation
criteria for COPD between 1 January 2014 and 31 owing to similar clinical presentations, and results
Received 10 November 2022 December 2018. Moderate exacerbation is defined will be interpreted with cautiousness.
Accepted 13 April 2023 ⇒ Although time-­fixed and time-­dependent confound-
as an outpatient visit and/or medication dispensation/
prescription for exacerbation; severe exacerbation is ers will be taken into account through modelling,
defined as hospitalisation for COPD. The primary outcomes unmeasured confounding is inevitable in observa-
of interest are the time to (1) first hospitalisation for a CV tional research, particularly when secondary data
event (including acute coronary syndrome, heart failure, are used.
arrhythmias or cerebral ischaemia) since cohort entry
or (2) death. Time-­dependent Cox proportional hazards
models will compare the hazard of a CV event between
exposed periods following exacerbation (split into these persistent respiratory symptoms and airflow
periods: 1–7, 8–14, 15–30, 31–180 and 181–365 days) limitation owing to airway and/or alveolar
and the unexposed reference time period, adjusted on abnormalities.2 The most common symp-
time-­fixed and time-­varying confounders. toms of COPD include dyspnoea, cough and
Ethics and dissemination Studies have been approved in sputum production; wheezing, chest tight-
Canada, Japan, the Netherlands, Spain and the UK, where
© Author(s) (or their ness and chest congestion are less common.3
an institutional review board is mandated. For each study,
employer(s)) 2023. Re-­use the results will be published in peer-­reviewed journals.
The disease may also involve acute worsening
permitted under CC BY-­NC. No of symptoms, referred to as COPD exacer-
commercial re-­use. See rights
and permissions. Published by
bation. COPD is associated with a twofold to
BMJ. INTRODUCTION threefold increased risk of having cardiovas-
For numbered affiliations see Chronic obstructive pulmonary disease cular disease (CVD), independent of age, sex,
end of article. (COPD) is a prevalent condition affecting smoking history or other confounders,4 and
approximately 10% of the adult population CVD contributes significantly to the burden
Correspondence to
Dr Clementine Nordon; globally1 and is a leading cause of prema- of disease in people living with COPD.4–8
​nordon.​clementine@g​ mail.​com ture death.2 The disease is characterised by Hence, the importance of identifying and

Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022 1

Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
monitoring cardiopulmonary risk factors or CVD in of COPD, compared with during the non-­exacerbation
patients with COPD is increasingly emphasised.9 10 periods.11 13–17 In a recent systematic review and meta-­
In addition to the strong interplay of COPD and analysis, the rate ratio of myocardial infarction in the first
CVD, an association between COPD exacerbation and 3 months after moderate or severe exacerbation of COPD
cardiovascular (CV) events is described in the litera- (compared with periods without exacerbation) was 2.4
ture.11–13 Specifically, the risk of experiencing an acute (95% CI 1.4 to 4.2), and that of ischaemic and/or haem-
CV event is increased by threefold to fivefold in the first orrhagic stroke was 1.7 (95% CI 1.2 to 2.4).18 Several
12 months following severe (hospitalised) exacerbation hypotheses have been suggested regarding the nature of

Table 1 Databases used for the EXACOS-­CV studies

Country Database Population and data used for the study
Canada Claims database: the Alberta ► Population covered: ≈4.3 million residents of Alberta province eligible for the
Health and Alberta Precision Alberta Health Care Insurance Plan
Laboratories44 built from ► Data sources: population registry; practitioner claims; National
the Alberta Health chronic Ambulatory Care Reporting System; pharmaceutical information; inpatient
disease cohort that identified hospitalisation—discharge abstract database; vital statistics
patients with COPD between
1 April 1985 and 31 March
2021
Spain Claims database: the BIG-­ ► Population covered: ≈1.9 million individuals who benefit from the public
PAC database45 national healthcare system, and followed up in various primary care centres
and hospitals of seven autonomous communities across Spain
► Data sources: outpatient visits (primary and specialist care); hospital and
emergency department data; outpatient and inpatient pharmacy dispensation;
date of death (registered by the GP)
Italy Claims database: Fondazione ► Population covered: ≈5 million residents of different local health units and
Ricerca e Salute46 47 regions in Italy
► Data sources: outpatient specialist visit; hospital data; drug dispensation:
all drugs reimbursed by the national healthcare service and supplied by
both local and hospital pharmacies; disease waiver claims for some chronic
diseases and death status (cause available when occurring in the hospital)
Germany Claims database: WIG226 ► Population covered: ≈4.5 million individuals from all parts of Germany and
using data from the German insured in one of the different sickness funds within the statutory health
statutory health insurance insurance system
system ► Data sources: demographics; outpatient care visits (reported for each
quarter of a year); pharmacy data; hospital data (emergency department: not
available); death status (cause available when occurring in the hospital)
The The PHARMO Database ► Population covered: ≈7 million individuals out of 17 million inhabitants of the
Netherlands Network-­validated Netherlands
algorithms48 in the ► Data sources: GP database; outpatient pharmacy database; hospital
Netherlands, a population-­ database—hospital admissions; date of death
based network of electronic
healthcare databases,
combine data from primary
and secondary healthcare
settings
The UK Electronic healthcare records: ► Population covered: ≈13 million individuals in England
CPRD Aurum linked with HES ► Data sources: GP database (Aurum); HES (hospital data, disease codes, dates
and National Statistics and procedures); death statistics
Japan Claims database: Medical ► Population covered: individuals taken care of in ≈22% of the 1727 ‘Diagnosis
Data Vision49 Procedure Combination’ hospitals in Japan (ie, hospitals that agreed to be
reimbursed by the government based on detailed healthcare utilisation—and
thus able to provide such information)
► Data sources: outpatient visits provided in the hospital; hospitalisation
(emergency department: not available); pharmacy dispensations; death status:
available when occurring in the hospital
COPD, chronic obstructive pulmonary disease; CPRD, Clinical Practice Research Datalink; EXACOS-­CV, EXAcerbations of COPD and their
OutcomeS on CardioVascular diseases; GP, general practitioner; HES, Hospital Episode Statistics.

2 Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022

 Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
the association between COPD exacerbation and a subse- To bridge these gaps, we have designed a series of
quent CV or cerebrovascular event (hereinafter referred retrospective longitudinal cohort studies that will be
to as CV events). The progressive inflammation and vessel conducted across multiple countries to evaluate the
wall abnormalities involved in COPD may lead to oxidative temporal association between exacerbation of COPD and
stress, arterial stiffness, ventilation–perfusion mismatch acute CV events requiring hospitalisation (hereinafter
and chronic hypoxaemia.19 In turn, an acute worsening referred to as severe CV events). Performing studies in
of airway and alveolar disease leading to hyperinflation multiple countries will allow us to explore the consistency
with additional hypoxaemia and increased cardiac stress and differences in results across healthcare systems, data-
may increase the risk of plaque rupture or of decompen- base types and patient populations.
sated heart failure (HF).20 21 The specific objectives of this study programme are (1)
Despite extensive research on the interplay between to determine the associations between periods following
CVD and COPD, important questions are still to be exacerbation of COPD and the occurrence of severe CV
addressed. First, there is conflicting evidence on whether events, including death, compared with periods without
or not exacerbation managed in the outpatient setting exacerbation; (2) to determine these associations for
(usually called moderate exacerbation) is also asso- moderate and severe exacerbation separately; and (3) to
ciated with an increased risk of CV events in the short characterise these associations after first, second or third
term.11 16 Portegies et al16 suggested the absence of such exacerbation of COPD in a subset of patients with an inci-
an association, whereas Rothnie et al17 and Reilev et al11 dent diagnosis of COPD.
suggested that moderate exacerbation is associated with
an increased risk of ischaemic stroke and myocardial
infarction. Second, the impact of the cumulative number
of previous exacerbations on the strength of association METHODS AND ANALYSIS
has only been explored in a single study that was limited Study design
to the 12 months preceding the CV event, which is a rela- The EXAcerbations of COPD and their OutcomeS on
tively short period of time, considering that COPD is a CardioVascular diseases (EXACOS-­ CV) Programme
chronic, progressive and incurable disease. The authors encompasses a group of longitudinal retrospective obser-
suggest that the greater the number of exacerbations, vational cohort studies that will analyse electronic health-
the weaker the association.17 This finding is counterin- care records or claims databases. Study locations will
tuitive because the cumulative number of exacerbations include Canada, Spain, Italy, Germany, the Netherlands,
over time is known to be associated with lung function the UK and Japan. Additional studies will be conducted in
decline,22 23 and poorer lung function is associated with a the USA, France and China; however, the current manu-
higher risk of CV events.24 25 Thus, it can be hypothesised script will not describe these three studies, since their
that the risk of experiencing a CV event increases with methodologies will be different, driven by the nature of
increasing number of COPD exacerbations. Finally, the data that can be accessed.
risk of a wide range of CV events, such as HF decompensa- The applicable, country-­ specific databases of
tion and new-­onset atrial fibrillation (AF) following exac- routinely collected electronic healthcare records or
erbation of COPD, has not been thoroughly explored. insurance claims to be used are described in table 1.
These elements are important from a clinical standpoint The selection period, eligibility criteria and definition
and could help physicians to identify more precisely of exposure and outcomes will be as comparable as
patients at risk of CV events and the periods of particular possible across all countries to the extent that infor-
risk of CV events, and in turn potentially improve quality mation is available. The study design and periods are
of care and patient management. detailed in figure 1.

Figure 1 Study design, selection period and follow-­up. CED, cohort entry date; COPD, chronic obstructive pulmonary disease;
CV, cardiovascular.

Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022 3

Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
Table 2 Definitions of COPD in the EXACOS-­CV studies
Country Identification of patients with COPD
27
Canada ► Diagnosis of COPD between 1 April 2014 and 31 March 2019, approximated by a medical encounter for COPD
(ICD-­9 codes 491, 492 and 496): ≥1 in the inpatient setting or ≥2 in the outpatient setting; the date of the first
medical encounter for COPD is the CED
► Age ≥40 years upon CED
Spain ► Diagnosis of COPD between 1 January 2014 and 31 December 2018, approximated by a medical encounter for
COPD (ICD-­10 codes J44, J44.0, J44.1, J44.8 and J44.9): ≥1 in the inpatient setting or ≥2 in the outpatient setting,
or, in case only 1 outpatient visit for COPD is identified, the presence of a spirometry measurement recorded in the
3 years prior or following the outpatient visit, the earliest code being the CED
► Age ≥40 years upon CED
Italy26 ► Diagnosis of COPD between 1 January 2015 and 31 December 2018, approximated by (1) criterion A: ≥1
hospitalisation for COPD (ICD-­9 codes 491.x, 492.x or 496 in a primary or secondary discharge code); (2) and/or,
if not available, criterion B: a disease waiver claim for COPD; (3) and/or, if not available, criterion C: ≥4 packs of
COPD-­related drugs dispensed in the same 12-­month period; CED is when the first criterion above is met
► Age ≥45 years upon CED
Germany28 ► Diagnosis of COPD between 1 January 2014 and 31 December 2018 approximated by ≥2 outpatient ICD-­10-­GM
diagnoses for COPD (as primary codes) in two different quarters of a year and/or ≥1 inpatient ICD-­10-­GM diagnosis
for COPD (primary or secondary codes) within the selection period; the date of the first medical encounter for
COPD is the CED
► Age ≥40 years upon CED
The ► Diagnosis of COPD between 1 January 2014 and 31 December 2018, approximated by an outpatient visit to the GP
Netherlands29 with a code for COPD (or free text)
► And (1) a GP-­reported GOLD classification, or (2) a spirometry measurement, regardless of result recorded in 3
years prior, or following the date of the first identified diagnosis of COPD (CED)
► Age ≥40 years upon CED
The UK30 ► Diagnosis of COPD using validated codes (SNOMED-­CT codes)
► Aged ≥40 years old at (the first) COPD diagnosis
► Current or ex-­smoker
Japan ► Diagnosis of COPD between 1 January 2015 and 31 December 2018, approximated by ≥2 outpatient codes for
COPD (ICD-­10 codes J41–44 in any position) or ≥1 inpatient discharge code for COPD (in any position) identified;
the earliest code will define the CED
► Age ≥40 years upon CED

CED, cohort entry date; COPD, chronic obstructive pulmonary disease; EXACOS-­CV, EXAcerbations of COPD and their OutcomeS on
CardioVascular diseases; GOLD, Global Initiative for Chronic Obstructive Pulmonary Disease; GP, general practitioner; ICD, International
Statistical Classification of Diseases; SNOMED-­CT, Systematized Nomenclature of Medicine—Clinical Terms.

Study population first severe CV event since CED, including death (first of any
In general, the study cohorts will include adult patients type, and first of each CV category), or (2) 31 December 2019
with COPD (using frequently used algorithm26–29 unless (administrative censoring), or (3) loss to follow-­up in the
another algorithm has been validated for the specific database (eg, owing to change of health insurance or leaving
database used30) who meet the following inclusion the medical practice). The follow-­up period stopped prior to
criteria: (1) at least two outpatient diagnosis codes for the start of the pandemic because of modifications in patient
COPD (International Statistical Classification of Diseases, behaviours and the patterns of diagnosis and management of
(ICD)-­10 code J44) or at least one inpatient diagnosis patients with COPD incurred by COVID-­19.31
code for COPD identified between 1 January 2014 and Regarding sample size, previous publications suggest
31 December 2018 (selection period); the first date of the possibility of identifying approximately 71 000 patients
COPD diagnosis identified during the selection period with COPD in Canada,32 45 000 in Spain,33 200 000 in
will define the cohort entry date (CED); (2) age ≥40 years Italy,26 250 000 in Germany,34 3600 in the Netherlands29
upon the CED. In addition, patients were to have contin- and 340 000 in the UK.35
uous data availability in the 24 months preceding the
CED. Specific details on COPD definition are provided in Definition of exposure
table 2. Patients with a record of alpha-­1 antitrypsin defi- The exposure of interest is acute exacerbation of COPD.
ciency (ICD-­10 E88.0) will be excluded. In turn, patients Definitions of moderate and severe exacerbation in each
will be considered as newly diagnosed with COPD, as country are provided in online supplemental table 1.
opposed to prevalent, if no diagnosis code for COPD is Moderate exacerbation of COPD will be generally
identified during the 24-­month baseline period. defined as (a) an outpatient visit to a physician for a
Patients with at least 1 day of follow-­up after CED will be reason related to COPD with (b) a prescription (approx-
included and will be followed from CED until the first of (1) a imated by a dispensation) of systemic corticosteroids

4 Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022

 Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
within the 5 days following the visit and for duration of exposure period will be a priori segmented into five mutu-
<15 days. Alternatives to this definition are listed in online ally exclusive time frames: 1–7 days, 8–14 days, 15–30 days,
supplemental table 1. The start date (day 1) of moderate 31–180 days and 181–365 days. For the study in the UK,
exacerbation of COPD is defined as the date of the outpa- an alternative data-­driven approach will be used, to iden-
tient visit or, in the absence of this information, the date tify time points at which the risk of a CV event changes
of systemic corticosteroid dispensation. following exacerbation. This approach may identify
Severe exacerbation of COPD will be defined as hospi- differences across CV outcomes, in terms of periods of
talisation of at least one night or a visit to the emergency time with greater rate of change.
department with a primary or secondary discharge code
for (acute exacerbation of) COPD or lower respiratory Outcome
tract infection (online supplemental table 1). The start The outcome of interest will be the time to the first non-­
date (day 1) of severe exacerbation according to primary fatal severe CV event or death (all-­cause, and then CVD or
discharge code will be the date of inpatient admission. COPD related). Non-­fatal severe CV events will comprise
Two exacerbations of COPD occurring within a 14-­day four categories: acute coronary syndrome (myocardial
period will be considered as the same event, and the infarction and unstable angina), HF decompensation
highest severity level will determine the overall severity. (including acute pulmonary oedema), cerebral isch-
aemia (ischaemic stroke and transient ischaemic attack)
Exposure periods and new or acute arrhythmias (AF and other arrhyth-
The exposure period during which a patient will mias, including resuscitated cardiac arrest). Non-­ fatal
contribute data on the risk of a severe CV event will be severe CV events will be defined as hospitalisation with a
the 365-­day period starting on day 1 of COPD exacer- primary or secondary discharge code related to the non-­
bation. This exposure period would last less than 365 fatal severe CV event(s) of interest (online supplemental
days in the case of a CV event, another exacerbation or table 1). Death will also be an outcome of interest and,
censoring. The time prior to exacerbation of COPD and when possible, will be categorised as CVD related, COPD
time beyond the end of a 365-­day exposure period will related or related to another reason. The first severe CV
be considered an unexposed period and will form the event will be categorised as (1) a CV event of any category
common reference period. All exacerbations of COPD (including non-­fatal events and death) and then (2) a CV
that occur after the CED will be measured, thus allowing event of each category, separately (eg, the first acute coro-
patients to contribute data for several exposure periods. nary syndrome, the first HF decompensation).
Several possible sequences of exposed and unexposed The date of a severe CV event will be determined using
periods in relation to patients are illustrated in figure 2. the same methodology as that for severe exacerbation (ie,
Based on previous studies, it is anticipated that the risk depending on whether the CV event is identified through a
of a severe CV event will not be constant over the 365-­ primary or a secondary discharge code). For severe CV events
day exposure period,13 17 and the highest risk is expected occurring during a hospital stay, the date of the event will be
to occur within the first 30 days.36 The post-­exacerbation determined using healthcare resource use in the hospital, for

Figure 2 Exposed and unexposed periods during the study. Patient A experiences two exacerbations of COPD and has a CV
event during an unexposed period; patient B experiences no exacerbation of COPD and has a CV event during an unexposed
period; patient C experiences one exacerbation during which a CV event occurs; patient D experiences no exacerbation of
COPD and is censored without an event. COPD, chronic obstructive pulmonary disease; CV, cardiovascular.

Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022 5

Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
Table 3 Potential confounders in the association between exacerbation of COPD and a severe CV event
Potential confounders Period of measurement
Age at cohort entry Baseline
Sex Baseline
Urban area Baseline
Year of cohort entry Baseline
Number of GP visits in the 12 months preceding cohort entry Time updated annually
Number of exacerbations since baseline period Time updated upon occurrence of
exacerbation
Newly diagnosed COPD (vs prevalent) Baseline
Season when AECOPD occurs (winter vs not) Time updated upon occurrence of
exacerbation
Comorbidities (ICD-­10 codes)
Smoking history (F17) Baseline
Alcohol use disorder (F10) Baseline
Obesity (E66) Baseline
37
Diabetes mellitus type 2 (E11) Baseline
Any disorders of lipoprotein metabolism and other dyslipidaemias,50 including any of the E78 codes (eg, Baseline
hyperlipidaemia (E78), hypercholesterolaemia (E78.0) and dyslipidaemia (E78.6, E78.1))
Ischaemic heart diseases (I20–I25) Baseline
Hypertensive diseases (I10–I15) Baseline
Heart failure (I50) Baseline
Pulmonary oedema (J81) Baseline
Pulmonary hypertension (I27.0, I27.2) Baseline
Venous thromboembolism (I26, I80, I82, I82, O08.2, O22.3, O87.1, O88.2) Baseline
Cerebrovascular disease (I60–I69) Baseline
Arrhythmia (I46, I48, I49) Baseline
Current asthma (J45/J46) Baseline
Chronic kidney disease and renal failure (N17–N19, I12.0, I13.1, I13.2) Baseline
Any severe mental illness,40 defined as recurrent and persistent depressive disorders (F33 and F34) and/ Baseline
or bipolar disorder (F30–31) and/or schizophrenia (F20–29)
Anxiety disorder (F40–48) Baseline
Medication (ATC codes)
Any use of cardiac agent (B01, C01–03, C07–09) Time updated annually
Any use of metabolic agent (C10 and A10) Time updated annually
Any use of COPD medication (R03: LABA, LAMA, ICS, SABA, SAMA, theophylline and roflumilast) Time updated annually

AECOPD, acute exacerbation of chronic obstructive pulmonary disease; ATC, Anatomical Therapeutic Chemical; COPD, chronic obstructive
pulmonary disease; CV, cardiovascular; GP, general practitioner; ICD, International Statistical Classification of Diseases; ICS, inhaled corticosteroids;
LABA, long-­acting beta 2-­agonist; LAMA, long-­acting muscarinic antagonist; SABA, short-­acting beta 2-­agonist; SAMA, short-­acting muscarinic
antagonist.

example, transfer to a stroke unit, a brain CT scan for stroke that end, the methodology will be broadly aligned across
or revascularisation for acute myocardial infarction. the seven countries to lend robustness and credibility
to the results. Missing data will be reported in terms of
Other measures percentages. The overall included patient cohort will
Potential confounders in the association between exacer-
be described via a Consolidated Standards of Reporting
bation of COPD and a severe CV event were determined
Trials flow diagram. General characteristics and poten-
based on the literature37–40 and confirmed based on
tial baseline confounders will be described overall and
discussions with subject matter experts (table 3).
separately for patients who ever or never had exacer-
Descriptive approach bation of COPD during the study period, as well as for
The analyses will follow a predetermined global statis- patients without any or at least one severe CV event
tical analysis plan, whereby country-­specific modifications during the study period. The same descriptive analyses
will be made according to the data source at hand. To will be conducted in the subgroup of patients who are

6 Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022

 Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
newly diagnosed with COPD. Details of follow-­ up will Ethics and dissemination
be reported, including summaries of total exposed and For all countries, except Germany and Italy, submission
unexposed time. The number and percentage of patients to an institutional review board (IRB) and approval are
with at least one severe CV event will be described, with necessary prior to data use. In Germany and Italy, as
severe CV events being described individually, by main per local regulation, the use of anonymised, secondary
category and overall. Kaplan-­ Meier curves will visually data does not require ethical approval. The studies have
summarise the cumulative incidence of a first CV event of already been approved in Canada ([Link]-­22-­0006,
any type and for each category. 3 March 2022), Japan (MINS-­REC-­220211, 7 April 2022),
the Netherlands (kenmerk AC/nWMO/22.013, 22 April
Modelling approach 2022), Spain (Ethics Committee of Consorci Sanitari de
Terrassa, protocol code AZ-­EXA-­2022-­06, 13 June 2022)
A time-­ to-­
event approach will be used, with the first
and the UK (approval obtained for the EXACOS-­ UK
severe CV event (including death) being the event of
Study35 East Midlands, Derby, REC reference number
interest. Because the hazard of a CV event is expected
05/MRE04/87). The general framework is as follows:
to change over the 365 days following day 1 of exacerba-
data repository or data providers (eg, Clinical Practice
tion, time-­dependent Cox proportional hazards models
Research Datalink) will provide data to the data analysis
will be used,41 with binary indicators of each exposed team under an ad hoc agreement. Data extraction will start
time frame, as time-­varying covariates. Although COPD after IRB agreement is confirmed. All data will be trans-
is a progressive disease, and the risk of CV events may ferred to the data analysis team after de-­identification.
increase overtime outside exacerbation periods, we For each individual country-­ based study, the results
decided to pool the unexposed periods, so as to simplify will lead to at least one publication in peer-­ reviewed
modelling and interpretation of results. However, if journals and an abstract submission to a conference. No
descriptive results reveal differences in crude incidence meta-­analysis of results is planned at this stage. Given the
rates of a CV event between the time period prior to first limited number of studies, it is not sure that statistical
exacerbation and the time period post-­365 days following power and precision of a random-­effects meta-­analysis
exacerbation, the reference unexposed time period may would be improved.
be revised prior to modelling, to include only time prior
to first exacerbation. Patient and public involvement
Time-­dependent Cox models will compare the hazard Although no patient was involved in the development of
of a first CV event between exposed subtime periods and the protocol, clinicians (pulmonologists and cardiolo-
the overall unexposed time period and will be fitted with gists) have been involved to take account of the clinical
and without adjustment for baseline and time-­ varying practice perspective and maximise the relevance of this
confounders. The target of inference will be the asso- research to patients.
ciation between each exposed subtime period (vs the
unexposed time period) and the outcome of time to first
severe acute CV event. For the first objective, analyses DISCUSSION
will be carried out for (1) time to first severe CV event To our knowledge, this programme of observational
of any category as the dependent variable and (2) for studies will be the first to explore the association between
separate categories of CV events as dependent variables. exacerbation of COPD and a wide range of CV outcomes
The latter will account for death as a competing risk by in routine clinical practice.43 The inclusion of newly diag-
nosed patients, followed up prospectively since the incept
modelling the cause-­specific hazard function.42 CV events
disease diagnosis, will provide an insight into patients’
of a different category may occur prior to the CV event of
trajectory. Additionally, the evaluation of associations
interest and will be handled as time-­varying confounders.
between exacerbation and CV events following the first,
When applicable, a sensitivity analysis will be conducted
second and third exacerbation of COPD will provide
in the newly diagnosed patients only in order to explore
important information on the impact of COPD progres-
any survival bias owing to the inclusion of prevalent sion on the strength of the relationship between COPD
patients. For objective 2, the model fitted for objective 1 and CVD.
will be refitted with an indicator of exacerbation severity Conducting the studies in countries with different
added to estimate specific associations for moderate and healthcare systems will also provide an insight into how
severe exacerbation. For objective 3, the analyses will be patterns of care, comorbidities and patient characteris-
conducted only in newly diagnosed patients (no diagnosis tics may differ while also checking the consistency of our
code for COPD identified during the 24-­month baseline results across countries.
period); the model fitted for objective 1 will be applied Several limitations apply. Having multiple databases
with indicators of the rank of the exacerbation added, with heterogeneous nature brings variability into the
to estimate separate associations for the first, second point estimate of associations evaluated. Namely, the
and third moderate or severe exacerbation during study precision and completeness of real-­ life diagnosis and
follow-­up. recording may vary between data sources. To minimise

Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022 7

 Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
the risk of measurement error, we have selected data- from GlaxoSmithKline, honoraria for educational events from AstraZeneca and
bases extensively used in medical research and applied Chiesi, and honoraria for participation on advisory board from Boehringer Ingelheim
and Chiesi, all paid to his institution. NMH has acted as speaker for and received
published validated algorithms and code lists where consulting fees from AstraZeneca, Novartis and Servier. EG has been chairwoman
feasible. Although CV events are the outcomes of interest, of a department at the Leibniz Institute that occasionally performed studies for
CV-­related death will not be used. Instead, all-­cause death pharmaceutical industries. The pharmaceutical companies included Byk-­Gulden,
will be measured. Secondary databases do not allow for a Nycomed, Bayer, Celgene, GlaxoSmithKline, Mundipharma, Novartis, Sanofi-­Aventis,
Sanofi Pasteur MSD and STADA. EG has been a consultant to Bayer-­Schering,
precise measure of the cause of death and adjudication Nycomed, GlaxoSmithKline, Schwabe, Teva, Novartis and AstraZeneca.
of the cause of death based on proxy variables is quite Patient and public involvement Patients and/or the public were not involved in
challenging. the design, or conduct, or reporting, or dissemination plans of this research.
This study should inform clinical practice on the timing Patient consent for publication Not required.
of interventions to reduce the risk of subsequent severe
Provenance and peer review Not commissioned; externally peer reviewed.
CV events among patients with COPD. In addition to
Supplemental material This content has been supplied by the author(s). It has
targeting the right patients at the right time with short-­ not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been
term strategies to reduce the risk of excess severe CV peer-­reviewed. Any opinions or recommendations discussed are solely those
events, long-­term preventive strategies that can reduce of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and
the risk of exacerbation of COPD should be considered. responsibility arising from any reliance placed on the content. Where the content
includes any translated material, BMJ does not warrant the accuracy and reliability
We hypothesise that reducing the risk of exacerbation, or of the translations (including but not limited to local regulations, clinical guidelines,
even preventing the first exacerbation, would reduce the terminology, drug names and drug dosages), and is not responsible for any error
overarching cardiopulmonary risk of patients, improving and/or omissions arising from translation and adaptation or otherwise.
their quality of life and their chances of longer-­ term Open access This is an open access article distributed in accordance with the
survival. Hence, preventive strategies that would reduce Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which
permits others to distribute, remix, adapt, build upon this work non-­commercially,
this cardiopulmonary risk may greatly improve the
and license their derivative works on different terms, provided the original work is
survival and quality of life of patients with COPD. properly cited, appropriate credit is given, any changes made indicated, and the use
We encourage decision-­making by health system admin- is non-­commercial. See: [Link]
istrators, clinicians and patients to consider the need for
ORCID iDs
multidisciplinary management of CV risk before, during Clementine Nordon [Link]
and following COPD exacerbation. Jennifer K Quint [Link]
Sami O Simons [Link]
Author affiliations Hana Müllerová [Link]
1
Medical and Payer Evidence, BioPharmaceuticals Medical, AstraZeneca,
Cambridge, UK
2
Respiratory Epidemiology, National Heart and Lung Institute, Imperial College
London, London, UK REFERENCES
3
Department of Medicine, Pulmonary and Critical Care Medicine, German Center for 1 Adeloye D, Song P, Zhu Y, et al. Global, regional, and national
Lung Research (DZL), Philipps University of Marburg, Marburg, Germany prevalence of, and risk factors for, chronic obstructive pulmonary
4
Department of Respiratory Medicine, NUTRIM School of Nutrition and Translational disease (COPD) in 2019: a systematic review and modelling analysis.
Lancet Respir Med 2022;10:447–58.
Research in Metabolism, Maastricht University Medical Centre, Maastricht, The 2 Global Initiative for Chronic Obstructive Lung Disease. Global
Netherlands strategy for the diagnosis, management and prevention of COPD;
5
Department of Medicine, University of British Columbia, Centre for Cardiovascular 2022.
Innovation, Vancouver, British Columbia, Canada 3 Miravitlles M, Ribera A. Understanding the impact of symptoms on
6
Clinical Epidemiology Group, Leibniz Institute for Prevention Research and the burden of COPD. Respir Res 2017;18:67.
4 Müllerova H, Agusti A, Erqou S, et al. Cardiovascular comorbidity in
Epidemiology—BIPS, Bremen, Germany COPD: systematic literature review. Chest 2013;144:1163–78.
5 Chen C-­C, Lin C-­H, Hao W-­R, et al. Association between chronic
Acknowledgements The authors acknowledge Narinder Bhalla for his help in the obstructive pulmonary disease and ventricular arrhythmia: a
protocol development. nationwide population-­based cohort study. NPJ Prim Care Respir
Med 2021;31:8.
Contributors CN and KR contributed to the development of the core protocol 6 Müllerová H, Marjenberg Z, Marshall J, et al. Risk of cardiovascular
and statistical analysis plan, and their adaptation in each country, and drafted the morbidity and mortality in people with chronic obstructive pulmonary
manuscript. JKQ, CFV, SOS and NMH contributed to the adaptation of the protocol disease versus those without COPD: a structured review of the
and statistical analysis plan in each country; they also contributed to drafting the evidence. Thorax 2021;76:A64–5.
manuscript and reviewed the last version. JM, MO, EG and HM contributed to the 7 Rabe KF, Hurst JR, Suissa S. Cardiovascular disease and COPD:
dangerous liaisons? Eur Respir Rev 2018;27:180057.
development of the core protocol and statistical analysis plan; they also contributed 8 Schneider C, Bothner U, Jick SS, et al. Chronic obstructive
to drafting the manuscript and reviewed its last version. pulmonary disease and the risk of cardiovascular diseases. Eur J
Funding This work is supported by AstraZeneca, the sponsor of the study, and is Epidemiol 2010;25:253–60.
conducted by external public or private research teams. 9 Simons SO, Elliott A, Sastry M, et al. Chronic obstructive pulmonary
disease and atrial fibrillation: an interdisciplinary perspective. Eur
Competing interests CN, KR, JM, MO and HM are employees of AstraZeneca and Heart J 2021;42:532–40.
own stock and/or stock options in the company. JKQ has received grants from MRC, 10 Hawkins NM, Peterson S, Ezzat AM, et al. Control of cardiovascular
HDR UK, GSK, Bayer, BI, asthma+lung, Chiesi and AstraZeneca; and personal fees risk factors in patients with chronic obstructive pulmonary disease.
for advisory board participation or speaking fees from GlaxoSmithKline, Boehringer Ann Am Thorac Soc 2022;19:1102–11.
11 Reilev M, Pottegård A, Lykkegaard J, et al. Increased risk of major
Ingelheim, AstraZeneca, Chiesi, Teva, Insmed and Bayer. CFV gave presentations adverse cardiac events following the onset of acute exacerbations of
at symposia and/or served on scientific advisory boards sponsored by Aerogen, COPD. Respirology 2019;24:1183–90.
AstraZeneca, Boehringer Ingelheim, CSL Behring, Chiesi, GlaxoSmithKline, Grifols, 12 Donaldson GC, Hurst JR, Smith CJ, et al. Increased risk of
Insmed, Menarini, Novartis, Nuvaira and MedUpdate. SOS has received grants myocardial infarction and stroke following exacerbation of COPD.
from Boehringer Ingelheim, AstraZeneca and GlaxoSmithKline, consulting fees Chest 2010;137:1091–7.

8 Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022

 Open access

BMJ Open: first published as 10.1136/bmjopen-2022-070022 on 26 April 2023. Downloaded from [Link] on November 16, 2024 by guest. Protected by copyright.
13 Kunisaki KM, Dransfield MT, Anderson JA, et al. Exacerbations of patients with COPD treated with single- or multiple-­inhaler triple
chronic obstructive pulmonary disease and cardiac events. A post therapy: a retrospective analysis in Spain. Chest 2022;162:1017–29.
hoc cohort analysis from the SUMMIT randomized clinical trial. Am J 34 Vogelmeier CF, Diesing J, Kossack N, et al. COPD exacerbation
Respir Crit Care Med 2018;198:51–7. history and impact on future exacerbations-­8-­year retrospective
14 Halpin DMG, Decramer M, Celli B, et al. Risk of nonlower respiratory observational database cohort study from Germany. Int J Chron
serious adverse events following COPD exacerbations in the 4-­year Obstruct Pulmon Dis 2021;16:2407–17.
UPLIFT® trial. Lung 2011;189:261–8. 35 Whittaker H, Rubino A, Müllerová H, et al. Frequency and severity of
15 Goto T, Shimada YJ, Faridi MK, et al. Incidence of acute exacerbations of COPD associated with future risk of exacerbations
cardiovascular event after acute exacerbation of COPD. J Gen Intern and mortality: a UK routine health care data study. Int J Chron
Med 2018;33:1461–8. Obstruct Pulmon Dis 2022;17:427–37.
16 Portegies MLP, Lahousse L, Joos GF, et al. Chronic obstructive 36 Hurst JR, Sin DD. Chronic obstructive pulmonary disease as a risk
pulmonary disease and the risk of stroke. The Rotterdam study. Am J factor for cardiovascular disease. A view from the Summit. Am J
Respir Crit Care Med 2016;193:251–8. Respir Crit Care Med 2018;198:2–4.
17 Rothnie KJ, Connell O, Müllerová H, et al. Myocardial infarction and 37 Castañ-Abad MT, Montserrat-­Capdevila J, Godoy P, et al. Diabetes
ischemic stroke after exacerbations of chronic obstructive pulmonary as a risk factor for severe exacerbation and death in patients
disease. Ann Am Thorac Soc 2018;15:935–46. with COPD: a prospective cohort study. Eur J Public Health
18 Müllerová H, Marshall J, de Nigris E, et al. Association of COPD 2020;30:822–7.
exacerbations and acute cardiovascular events: a systematic 38 Dong H, Hao Y, Li D, et al. Risk factors for acute exacerbation of
review and meta-­analysis [In press]. Ther Adv Respir Dis chronic obstructive pulmonary disease in industrial regions of China:
2022;16:17534666221113647. a multicenter cross-­sectional study. Int J Chron Obstruct Pulmon Dis
19 Morgan AD, Zakeri R, Quint JK. Defining the relationship between 2020;15:2249–56.
COPD and CVD: what are the implications for clinical practice? Ther 39 Njoku CM, Alqahtani JS, Wimmer BC, et al. Risk factors and
Adv Respir Dis 2018;12:175346581775052. associated outcomes of hospital readmission in COPD: a systematic
20 Čelutkienė J, Balčiūnas M, Kablučko D, et al. Challenges of treating review. Respir Med 2020;173:105988.
acute heart failure in patients with chronic obstructive pulmonary 40 Correll CU, Solmi M, Veronese N, et al. Prevalence, incidence
disease. Card Fail Rev 2017;3:56–61. and mortality from cardiovascular disease in patients with pooled
21 Alter P, van de Sand K, Nell C, et al. Airflow limitation in COPD is and specific severe mental illness: a large-­scale meta-­analysis of
associated with increased left ventricular wall stress in coincident 3,211,768 patients and 113,383,368 controls. World Psychiatry
heart failure. Respir Med 2015;109:1131–7. 2017;16:163–80.
22 Donaldson GC, Seemungal TAR, Bhowmik A, et al. Relationship 41 Zhang Z, Reinikainen J, Adeleke KA, et al. Time-­varying covariates
between exacerbation frequency and lung function decline in chronic and coefficients in COX regression models. Ann Transl Med
obstructive pulmonary disease. Thorax 2002;57:847–52. 2018;6:121.
23 Dransfield MT, Kunisaki KM, Strand MJ, et al. Acute exacerbations 42 Austin PC, Lee DS, Fine JP. Introduction to the analysis of survival
and lung function loss in smokers with and without chronic data in the presence of competing risks. Circulation 2016;133:601–9.
obstructive pulmonary disease. Am J Respir Crit Care Med 43 Dransfield MT, Criner GJ, Halpin DMG, et al. Time-­dependent risk
2017;195:324–30. of cardiovascular events following an exacerbation in patients with
24 Silvestre OM, Nadruz W, Querejeta Roca G, et al. Declining lung chronic obstructive pulmonary disease: post hoc analysis from the
function and cardiovascular risk: the ARIC study. J Am Coll Cardiol impact trial. J Am Heart Assoc 2022;11:e024350.
2018;72:1109–22. 44 Chen G, Farris MS, Cowling T, et al. Prevalence of atherosclerotic
25 Wang B, Zhou Y, Xiao L, et al. Association of lung function with cardiovascular disease and subsequent major adverse
cardiovascular risk: a cohort study. Respir Res 2018;19:214. cardiovascular events in Alberta, Canada: a real-­world evidence
26 Calabria S, Ronconi G, Dondi L, et al. Open triple therapy for study. Clin Cardiol 2021;44:1613–20.
chronic obstructive pulmonary disease: patterns of prescription, 45 Sicras-­Mainar A, Enriquez JL, Hernández I, et al. PMU146 validation
exacerbations and healthcare costs from a large Italian claims and representativeness of the Spanish BIG-­PAC database: integrated
database. Pulm Pharmacol Ther 2020;61:101904. computerized medical records for research into epidemiology,
27 Gershon AS, Wang C, Guan J, et al. Identifying individuals with medicines and health resource use (real word evidence). Value in
physcian diagnosed COPD in health administrative databases. Health 2019;22:S734.
COPD 2009;6:388–94. 46 Calabria S, Dondi L, Ronconi G, et al. Acute lower respiratory
28 Mueller S, Gottschalk F, Groth A, et al. Primary data, claims data, and infections: real-­world evidence of antibiotic prescription pattern
linked data in observational research: the case of COPD in Germany. and costs from a large administrative Italian database. Fam Pract
Respir Res 2018;19:161. 2022;39:669–77.
29 Overbeek JA, Penning-­van Beest FJA, Balp M-­M, et al. Burden of 47 Calabria S, Ronconi G, Dondi L, et al. Coronary artery disease in
exacerbations in patients with moderate to very severe COPD in the patients older than 35 and eligible for cardiovascular secondary
Netherlands: a real-­life study. COPD 2015;12:132–43. prevention: an Italian retrospective observational analysis of
30 Quint JK, Müllerova H, DiSantostefano RL, et al. Validation of chronic healthcare administrative databases. J Clin Med 2021;10:4708.
obstructive pulmonary disease recording in the clinical practice 48 Herings RMC, Pedersen L. Pharmacy-­based medical record
research Datalink (CPRD-­GOLD). BMJ Open 2014;4:e005540. linkage systems. In: Strom BL, Kimmel SE, Henness S, eds.
31 Simons SO, Hurst JR, Miravitlles M, et al. Caring for patients with Pharmacoepidemiology. 5th ed. John Wiley & Sons, Ltd,
COPD and COVID-­19: a viewpoint to spark discussion. Thorax 2012: 270–86.
2020;75:1035–9. 49 Laurent T, Simeone J, Kuwatsuru R, et al. Context and
32 Tran DT, Thanh NX, Ohinmaa A, et al. Current and future direct considerations for use of two Japanese real-­world databases in
healthcare cost burden of chronic obstructive pulmonary disease in Japan: medical data vision and Japanese medical data center. Drugs
Alberta, Canada. Can J Respir Crit Care Sleep Med 2020;4:39–47. - Real World Outcomes 2022;9:175–87.
33 Alcázar-­Navarrete B, Jamart L, Sánchez-­Covisa J, et al. Clinical 50 Wang M-­T, Lo Y-­W, Tsai C-­L, et al. Statin use and risk of COPD
characteristics, treatment persistence, and outcomes among exacerbation requiring hospitalization. Am J Med 2013;126:598–606.

Nordon C, et al. BMJ Open 2023;13:e070022. doi:10.1136/bmjopen-2022-070022 9