Article Text
PDF
Abstract
Introduction Chronic obstructive pulmonary disease (COPD) is one of the leading causes of death worldwide, which is partly contributed to the increasing prevalence of COPD owning to a demographic shift towards an older population. Conversely, recent studies on COPD mortality that take this demographic shift in age into account find decreasing overall age-standardised COPD mortality rates over time. This decrease in the age-standardised COPD mortality rate is contributed advances in COPD diagnostics and treatment, decreasing smoking prevalence and general advances in medical care particularly in western countries. However, it is unknown if patients with COPD have experienced a comparable relative increase in survival in line with the general population.
Hence, there is a need for longitudinal studies comparing trends in mortality in patients with COPD compared with matched non-COPD individuals from the background population.
Methods and analyses This is a cohort study with a matched non-COPD comparator cohort. Data are retrieved from the Danish national registers. Data from multiple registries from 1983 to 2018 will be merged on an individual level using the 10-digit Civil Registration numbers that are unique to each citizen in Denmark. Time trends in mortality in patients with COPD compared with the matched comparator cohort will be examined in three study periods: 1983–1993, 1994–2007 and 2008–2018.
Ethics and dissemination The study is entirely based on registry data and ethical approval is not required according to Danish Law and National Ethics Committee Guidelines. The results will be published in peer-reviewed journals and reported at appropriate national and international conferences.
- pulmonary disease, chronic obstructive
- epidemiologic studies
- risk factors
- mortality
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
This study is conducted on a large population-based cohort. The registries cover the entire Danish population, which ensures immense amounts of data, resulting in great statistical power and limited selection bias.
The study will include a comparison cohort from the general Danish population, matched by year of birth, sex, cohabitation status and residence.
The study design allows for the exploration of changes in mortality over time in patients with chronic obstructive pulmonary disease compared with matched comparators.
We expect to have complete follow-up on all cases and comparators.
A limitation of register-based studies is potential confounding. We will be able to control for many important potential confounders, but residual confounding is likely to affect our results.
Introduction
Chronic obstructive pulmonary disease (COPD) is one of the leading causes of morbidity and mortality globally.1 COPD is often diagnosed late in life due to its slow progressive nature and the prevalence of COPD is expected to increase in the future mainly because of the high prevalence of tobacco smoking and increasing life expectancies in the general population.2
Life expectancy and standard of living have increased in industrialised countries during the past century. In Denmark, the life expectancy increased from 50 to 80 years from 1900 to 2018 largely due to high standards of medical care and universal free tax-funded healthcare for all citizens.3 However, despite the overall improvement in health and survival and an increased focus on COPD diagnosis and preventative measures, COPD remains the leading global cause of disability and mortality.4 In Denmark, 16% of patients with COPD who are hospitalised die within 30 days of discharge.5
Our current knowledge about the development of COPD mortality during the past decades is mainly based on data from death certificate reporting on causes of death or underlying causes of death.6 Recent studies on COPD mortality that take the demographic shift towards an older population into account find that the overall age-standardised COPD mortality rate is decreasing over time across the world. In most countries, the largest decrease is seen for men whereas a stable or increasing age-standardised mortality rate is seen for women.6–12
Mortality studies based on death certificates describe the burden of deaths related to COPD. However, studies show, that COPD is often under-reported as an underlying cause of death13–15 leading to an underestimation of mortality related to COPD. Also, coding variations among coders over time and coding changes between and within International Classification of Disease (ICD) revisions may hamper comparison of data on causes of death over long time periods.16 Moreover, these cross-sectional studies lack the ability to provide information on mortality in patients with COPD in relation to the time of diagnosis. Cohort studies investigating mortality in patients with COPD (ie, including patients with COPD who die from other causes than COPD) are lacking. To our knowledge, no previous studies have compared the time trends in COPD mortality to the increasing life expectancy of the general population. It is unknown if patients with COPD have experienced a comparable relative increase in survival in line with the general population.
The aim of this nationwide cohort study with a matched comparator cohort is to investigate the over all mortality at an individual level from the time of COPD diagnosis and to investigate the time trends in over all mortality in patients with COPD compared with a matched comparator cohort from 1983 to 2018.
Methods and analysis
Study design
This nationwide observational registry-based cohort study is conducted in Denmark. Prospectively collected data from national administrative registries17 will be merged using the unique 10-digit Danish Civil Registration (CPR) number.18 The CPR number facilitates exact, individual-level linkage of all the national databases.
Data sources
Several national registries will be used in the study.
The Danish Civil Registration System (DCRS) has recorded information on date of birth, sex, residency, cohabitation status and daily electronic updates on migration and vital status since 1968. Data from the DCRS are considered precise and complete.18
The Danish National Patient Registry (DNPR)19 collects data on diagnoses, diagnostic procedures and administrative information from all Danish hospitals. The DNPR was initiated in 1977 and achieved complete nationwide coverage of all non-psychiatric admissions since 1978 and outpatient clinic, emergency department contacts and hospital psychiatric contacts since 1995. Primary and secondary diagnoses are classified according to the coding version ICD-8 (International Classification of Diseases and Related Health Problems 8th Revision) until 1993 and the ICD-10 Classification version since 1994 and onwards.
The Danish Register of Causes of Death20 holds information on causes of death since 1970.
Statistics Denmark (DTS)21 is an administrative database that collects individual-level data on socioeconomic factors.
Study population
Copd case population
Patients who are ≥40 years of age at the time of diagnosis with a first-time primary or secondary diagnostic code registration of COPD in the DNPR between 1 January 1983 and 31 December 2018 are included in the case population. Primary and secondary diagnoses are classified according to the coding version ICD-8 until 1993 and the ICD-10 Classification version since 1994 and onwards.19 The positive predictive value (PPV) for COPD hospitalisation in the DNPR is 92% (95% CI 91% to 93%).22 COPD definitions according to the two ICD versions can be seen in table 1.
Overview of the ICD-8 and ICD-10 diagnostic codes used to define COPD and the study periods for when they are used
Matched comparator cohort
A randomly selected comparison cohort without COPD is identified from the background population using the DCRS. Each COPD case is matched with four comparators by birth year, gender, cohabitation status and residency. We use a matched cohort study design because we are then able to account for several important confounders. The matching will also to some degree account for important unmeasured confounders (eg, air pollution, changes in public healthcare, better living conditions) that are likely to change over time. Hence, it is the relative differences in mortality between COPD cases and comparators and the change over time in this difference that is investigated in this matched cohort study.
Study period
Cases and comparators are followed until death, migration or end of follow-up in 2018, whichever comes first. Cases are included from 1983 to 2018. To ensure, that cases are included at the time of their first COPD-related hospital contact, a 6-year wash-out period is applied (1977–1982) excluding patients diagnosed before the inclusion period.
The study population (cases and comparators) is divided into three study periods in order to evaluate changes over time of the outcomes (figure 1). Cases and comparators are included in the study period that includes the year of diagnosis. For example, if a case is diagnosed in 1989, the case and matched comparators are included and followed during study period 1.
Illustration of the study period, inclusion periods, wash-out period, end of follow-up and important COPD and smoking-related landmarks. COPD, chronic obstructive pulmonary disease; ICD, International Classification of Disease; LAMA, long-acting muscarinic antagonist.
The study periods are defined as follows:
Period 1: 1983–1993.
Period 2: 1994–2007.
Period 3: 2008–2018.
The three periods were defined based on important COPD and smoking-related landmarks and ICD coding versions. To uniform the definitions of COPD in the periods we defined period 1 based on the period where ICD-8 coding was used. In period 2 and period 3, COPD was defined by ICD-10 codes. During period 2, the first GOLD recommendations (2001) and revision (2006) were published. Also, around year 2000, long-acting muscarinic antagonist inhalers were introduced in Denmark. An important landmark against smoking was established in 2007 (period 3) with an antismoking law that prohibited smoking in most public spaces. Also, a fast-track cancer pathway for lung cancer was introduced in 2008, which is likely to have positively influenced the survival of patients with COPD.
Outcomes
The primary outcome is all-cause mortality. Information on the date of death is obtained from the DCRS. The secondary outcome is causes of death for cases and comparators in the three study periods.
Baseline characteristics
Information on baseline characteristics for cases and comparators, including the registries from which they are obtained, is displayed in table 2. Information on age, sex and municipality of residency is obtained from the DCRS. Information on cohabitation status is obtained from DTS.
Overview of baseline variables and covariates included in the study
Confounding
Information on potential confounders between the exposure (COPD) and outcome (death) was sought in the literature. Potential confounders can be seen in the directed acyclic graph23 (figure 2). Age, sex, cohabitation status and municipality of residency will be controlled for by matching with the comparators. Comorbidities and income will be adjusted for in the cox regression analysis. Information on comorbidities (ICD-10 diagnostic codes) 3 years prior to diagnosis is obtained from the DNPR to estimate the COPD-specific Comorbidity Test (COTE index)24 and the Charlson Comorbidity Index (CCI)25 (table 2). The COTE index is a point scale index evaluating COPD-specific comorbidities and is a predictor of risk of death in COPD. The CCI is a weighted score, which indicates the disease burden and is a prognostic indicator of mortality. The PPV of CCI in the DNPR is 98% (95% CI 96.9% to 98.8%).26 Information on cohabitation status and income will be obtained from DTS (table 2). In Denmark, all citizens receive a government-funded age pension from approximately the age of 65 years. Income will, therefore, only include the study population between 40 and 64 years. Unmeasured confounding by smoking will be addressed by using the E-value method27 and information on smoking prevalence in the general population in the three time periods will be obtained from DTS. Information on the highest obtained educational level and occupational exposures is not available for the whole study period, and therefore, not included in the analyses. Unmeasured confounding by air pollution will partly be accounted for by matching on the place of residency.
Directed acyclic graph illustrating the assumed causal relationships among the variates and covariates in the current study. COPD, chronic obstructive pulmonary disease. Exposure: COPD. Outcome: death. White circle: adjusted variable. Grey circle: other variable. Green circle: ancestor of outcome. Green arrow: causal path. Pink arrow: biasing path.
Statistical analysis plan
Baseline characteristics will be presented in tables as proportions, except age, which is summarised as a mean with SD. Differences in income and CCI between cases and comparators for each study period is estimated using the χ2 test. Survival will be estimated for each study period using the Kaplan-Meier method. HRs will be estimated using the Cox proportional hazard model. If the HR is non-proportional, other statistical models will be considered. The HR for mortality will be adjusted for income and CCI. Mortality will be stratified by gender. A difference-in-difference analysis28 will be used to test whether the difference in mortality (HR) between case and comparators is significantly different between the three time periods (ie, changes over time). We do not have information on smoking status. We will use the E-value method27 to perform sensitivity analysis on smoking as a potential confounder. Causes of death will be presented as proportions. The most common causes of death for cases and comparators will be ranked and displayed in a table or bar chart. Estimates will be provided with 95% CIs and a significance level of 0.05. Data management and statistical analyses are performed in SAS V.9.4 TS Level 1M5 (SAS). SAS SURVEYSELECT procedure is used for selecting probability-based random comparison cohort.
Discussion and perspectives
Updated mortality trends in COPD provide insight for targeted interventions. There are some limitations to the study. The COPD diagnosis is based on the treating physician coding COPD when the patients are in contact with the hospital. Hence, we do not know if the diagnose has, for example, been confirmed with spirometry. The definition of COPD in period 1 is based on ICD-8 codes for emphysema, bronchitis and chronic bronchitis which may misclassify some cases. Hence, the comparison of period 1 to the remaining periods must take this difference in coding and risk of misclassification into account.
We will be able to control for many important potential confounders by matching and adjusting. The matched comparison cohort is exposed to the same air pollution, changes in public healthcare, etc as the COPD cohort. Hence, many of these unmeasured confounders are, therefore, taken into account by the matching. Nevertheless, residual confounding is likely to affect our results which must be taken into account when interpreting the results of the study.
We expect to find an increase in COPD mortality when compared with the background population.
We expect these novel findings of the negative trends in COPD mortality to motivate more research on the topic as well as an increased focus from healthcare policy-makers in allocating more resources to improve early diagnosis and other preventative measures.
Ethics
The study is entirely register-based, and there are no human participants. Informed consent or research ethics approval is, therefore, not required according to Danish Law and National Ethics Committee Guidelines.
Patient and public involvement
Patients or the public were not involved in the development of the design or dissemination plans of our research.
Storage and management of data
Data are anonymised from the national databases. Raw data have been obtained from and is stored at DST. Data management of the raw data and some baseline characteristics have been computed. Statistical analyses will commence November 2023. All analyses will be carried out using a secure remote access to a virtual desktop at DST. Data management has been carried out by experienced statisticians from i2minds.
Dissemination of results
The results of this study will be published in international peer-review journals. The results will also be presented at relevant national and international meetings.
Ethics statements
Patient consent for publication
References
Footnotes
Contributors All authors have contributed substantially to the conception and design of this study protocol. MGS, AL and OH have conceptualised the research question. MGS, AL, OH and RI have collaborated on designing the study. RI was lead on describing the statistical analyses plan. MGS wrote the first draft of the protocol with input from AL, OH and RI. All authors have revised the manuscript.
Funding This work is supported by an unrestricted grant from The Danish Lung Association Foundation and the Eva Merete Falck Crone Foundation. Award/grant number is not applicable.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.