This is a retrospective cohort study comparing COPD patients aged 40 years and older with new exposure to roflumilast with up to 5 unexposed (ie, not roflumilast-exposed) COPD controls matched by propensity score (PS), age, sex, and year of cohort entry. The study is using electronic healthcare databases in the US (Military Health System database), Germany (GER) (German Pharmacoepidemiological Research Database), and Sweden (SWE) (national databases including healthcare, death, and demographics data). The main objective of this study is to evaluate the long-term safety of roflumilast in the treatment of COPD with main focus on the primary outcome of all-cause mortality and evaluation of potential safety issues identified during the clinical trials of roflumilast. Crude mortality and incidence rates will be compared between roflumilast exposed and unexposed COPD patients and adjusted hazard ratios (HRs) of the primary and secondary outcomes will be calculated.

Rationale and background Roflumilast is licensed (i) for maintenance treatment of severe chronic obstructive pulmonary disease (COPD) (FEV1 post-bronchodilator less than 50% predicted) associated with chronic bronchitis (CB) in adult patients with a history of frequent exacerbations as add on to bronchodilator treatment (EU) and (ii) as a treatment to reduce the risk of COPD exacerbations in patients with severe COPD associated with chronic bronchitis and a history of exacerbations (US). As a condition of approval for marketing in the EU, the European Medicines Agency requested the Marketing Authorisation Holder to conduct a long-term comparative observational safety study. Research question and objectives The main objective of this study is to evaluate the long-term safety of roflumilast in the treatment of COPD with main focus on 5-year all-cause mortality and evaluation of potential safety issues identified during the development programme of roflumilast. Study design Retrospective cohort study comparing COPD patients aged 40 years and older with new exposure to roflumilast with up to 5 unexposed (i.e. not roflumilast exposed) COPD controls matched by propensity score (PS), age, sex, and year of cohort entry. PS matching was used as an established method to control for confounding. Crude mortality and incidence rates were compared between roflumilast exposed and unexposed COPD patients and adjusted hazard ratios (HRs) of the primary and secondary outcomes were calculated in GER and SWE, while in the US adjusted HRs were only calculated for the primary outcome and are not yet available for secondary outcomes. Reported adjusted HRs for secondary outcomes therefore only refer to GER and SWE. Setting Electronic healthcare databases in the United States (US), Germany (GER), and Sweden (SWE) holding demographic data, data on health, and dispensing of medication. The initial assessment of the number of cohorts in each database including the years 2011, 2012, and 2013 resulted in a total of 54985 (9573 exposed), 50493 (8775 exposed), and 18602 (3207 exposed) COPD patients in the US, GER, and SWE, respectively. Variables and data sources Exposure variable: Use of roflumilast (ATC code R03DX07) Outcomes: The primary outcome in the study is 5-year all-cause mortality. Secondary outcomes are death by suicide or hospitalisation for suicide attempt, hospitalisation for any cause, major cardiovascular events, respiratory disease related hospitalisation, new diagnosis of depression, new diagnosis of malignant neoplasm, hospitalisation due to serious diarrhoea of non-infectious origin, abnormal and unexplained weight loss, and new diagnosis of tuberculosis or hepatitis B or C or other severe viral hepatitis infection (except hepatitis A). Other covariates: Characterisation of baseline therapy, baseline medical history and other socio-demographic covariates. Data sources: Military Health System database (US), German Pharmacoepidemiological Research Database (GER), national databases including healthcare, death, and demographics data (SWE).

Exposed Cohort selection

Inclusion Criteria:

• Have one or more prescription(s) of roflumilast with the date of first prescription/dispensation defined as the Cohort Entry Date
• Have active data in the respective database(s) for at least 1 year prior to the Cohort Entry Date
• Be at least 40 years old at Cohort Entry Date
• Have a diagnosis of COPD or chronic bronchitis prior to the Cohort Entry Date (databases with outpatient diagnoses) or (databases without outpatient diagnoses) either a hospital diagnosis of COPD or chronic bronchitis prior to the Cohort Entry Date or a proxy for such indication based on prescription data.

Non-Exposed Cohort selection Patients meeting the following inclusion criteria belong in the Non-Exposed Cohort and are eligible as Matched Non-Exposed controls. Each patient in the Non-exposed Cohort selected as matched exposed patient is assigned as a Cohort Entry Date the date of the first roflumilast prescription of the matched exposed patient.

unselected COPD patients aged 40 years and older, reflecting the use of roflumilast in a real-life setting.