A Randomised Effectiveness Study Comparing Fluticasone Furoate (FF, GW685698)/Vilanterol (VI, GW642444) With Standard Treatment in Chronic Obstructive Pulmonary Disease (COPD)

Incidence of SAE of pneumonia was defined for each randomized treatment group as the proportion (number) of participants in that group who experienced at least one SAE of pneumonia in the Pneumonia Adverse Event of Special Interest subgroup during the treatment period (from start date of exposure to stop date of exposure + 28 days). Non-inferiority is demonstrated if the upper limit of the two-sided 95% confidence interval for the incidence ratio is less than 2. Serious Adverse Events of pneumonia are defined by the Pneumonia Special Interest Group, which for SAEs of pneumonia collected for these subjects includes the following preferred terms: Pneumonia, Pneumonia aspiration, Aspergillus infection, Empyema, Pneumonia streptococcal, Pneumonitis, Pulmonary tuberculosis.

The mean number of SAE of pneumonia over the treatment period (from first date of exposure to last date of exposure + 28 days was calculated. Analysis was performed using a negative binomial regression model with covariates of randomised treatment and with logarithm of time on treatment as an offset variable.

The analysis method was a Cox proportional hazards model adjusted for randomized treatment. Analyses included those on-treatment SAEs of pneumonia that had an onset over the first 364 days of exposure, as defined. Participants who did not have an SAE of pneumonia during the first 364 days of the treatment period (start date of exposure to end date of exposure + 28 days were considered censored. Number of participants with event is presented.

A COPD-related secondary care contact was defined as an inpatient admission or a specialist outpatient visit or an accident \& emergency (A\&E) contact. A participant with an A\&E contact and subsequent inpatient admission was considered to have had two healthcare contacts. Inpatient admissions recorded at two hospitals on the same day, this was counted as a single (inpatient admission) secondary care contact. COPD-related contacts were identified using predefined lists of ICD-10 codes, specialty descriptions and diagnosis codes recorded in the patients electronic health record (EHR). GLM assuming the negative binomial distribution with log-link function and logarithm of time on treatment as an offset variable and adjusted for randomised treatment, baseline COPD maintenance therapy per randomisation stratification, number of moderate/severe COPD exacerbations in the previous year to randomisation and smoking status at baseline.

A COPD-related primary care contact was defined as a primary care contact on a given calendar date with either a nurse, general physician (GP) or other healthcare professional that were considered as COPD-related, if the most prominent signs and symptoms the participant was presenting were as a direct result of the participant's COPD, as per Readcodes recorded in the patients electronic health record (EHR). The analysis method was General Linear Model assuming an underlying negative binomial distribution with a log-link function and logarithm of time on treatment as an offset variable and adjusted for randomised treatment, baseline COPD maintenance therapy per randomisation stratification, number of moderate/severe COPD exacerbations in the previous year to randomisation and smoking status at baseline.

A secondary care contact was defined as an inpatient admission or a specialist outpatient visit or an A\&E contact. A participant with an A\&E contact and subsequent inpatient admission was considered to have had two healthcare contacts. In the situation where inpatient admissions were recorded at two hospitals on the same day, this was counted as a single secondary care contact. The analysis method was General Linear Model assuming an underlying negative binomial distribution with a log-link function and logarithm of time on treatment as an offset variable and adjusted for randomised treatment, baseline COPD maintenance therapy per randomization stratification, number of moderate/severe COPD exacerbations in the previous year to randomization and smoking status at baseline.

A primary care contact was defined as contact with a either a nurse, general practitioner, or other healthcare professional. The analysis method was General Linear Model assuming an underlying negative binomial distribution with a log-link function and logarithm of time on treatment as an offset variable and adjusted for randomised treatment, baseline COPD maintenance therapy per randomisation stratification, number of moderate/severe COPD exacerbations in the previous year to randomisation and smoking status at baseline.

Initial therapy was defined as the treatment that the subject was randomised to at randomisation. Discontinuation of initial therapy was defined as any modification of initial therapy. These included stepping up, stepping down or switching to another class/class combination, or withdrawal from the study. Switching within the same drug class did not count unless participant switched from FF/VI to a different ICS/LABA. The analysis method was a Cox proportional hazards model adjusted for randomized treatment. The probability of discontinuation of initial therapy was measured from the date of randomisation (i.e., exposure start date) to the date of discontinuation of initial therapy to which the participant was randomized, or date of treatment termination (Visit 6 or early withdrawal visit) for participants who completed the study without discontinuing the initial therapy (censored). Number of participants with event is presented.

The date of an event for addition of a further COPD controller medication was defined as the exposure start date of the first modified treatment medication that included a new COPD maintenance therapy of a new class of drug (to the initial therapy) during the study treatment period, as collected on the investigational product page of the eCRF. Participants who did not add any COPD controller medication during the study were censored at the end of the treatment period (Day 364). This was equivalent to stepping up, defined as the addition of at least one new class of drug. The probability of an event was measured from the date of randomisation (i.e., treatment initiation) to the date of a change event. The analysis method was a Cox proportional hazards model adjusted for randomized treatment. Number of participants with event is presented.

The date of an event for moderate / severe COPD exacerbation was defined as the exacerbation onset date. The analysis method was a Cox proportional hazards model adjusted for randomized treatment. The probability of a first moderate / severe exacerbation was measured from the date of randomisation (i.e., treatment initiation) to the onset date of first moderate or severe COPD exacerbation, as recorded on eCRF, or date of treatment termination (Visit 6 or early withdrawal visit) for participants who completed the study without any moderate or severe exacerbations (censored). Participants who completed the study without a moderate or severe COPD exacerbation and analyses of time to first moderate/severe exacerbation were censored at Day 364. Number of participants with event is presented.

The date of an event for moderate / severe COPD exacerbation was defined as the exacerbation onset date. The analysis method was a Cox proportional hazards model adjusted for randomized treatment. The probability of first moderate / severe exacerbation on initial therapy was measured from the date of randomisation (i.e., exposure start date) to the onset date of first moderate or severe COPD exacerbation, or to the date of discontinuation of initial therapy (analysis was censored at date of discontinuation of initial therapy) for participants who completed the study without any moderate or severe exacerbations on initial therapy. Number of participants with event is presented.

The date of an event for severe exacerbation was defined as the exacerbation onset date. Participants who completed the study without a severe exacerbation were censored. The analysis method was a Cox proportional hazards model adjusted for randomized treatment.

The probability of first severe exacerbation was measured from the date of randomisation (i.e., treatment initiation) to the onset date of first severe exacerbation, as recorded on eCRF, or date of treatment termination (Visit 6 or early withdrawal visit) for participants who completed the study without any severe exacerbations (censored). At Day 364, all participants who have not experienced a severe exacerbation are considered censored, regardless of whether their on-treatment phase continues beyond day 364, including those who withdrew early. Number of participants with event is presented.

All SAEs included in the AE subgroup of special interest of "pneumonia" were considered as an SAE of pneumonias. A fatal SAE was defined as a SAE with outcome of fatal for study participant. The number of participants with fatal SAEs of pneumonia was assessed over 14 months.

The number of participants with non-serious ADRs was assessed for up to 54 weeks. An ADR is any untoward medical occurrence in a patient or clinical investigation subject, temporally associated with the use of a medicinal product, for which there is a reasonable possibility that the untoward occurrence is causally related to the medicinal product. A non-serious ADR included one of the following: exacerbation of chronic or intermittent pre-existing condition; signs, symptoms, or the clinical sequelae of a suspected interaction; signs, symptoms, or new conditions detected or diagnosed after study treatment administration.

SAEs assessed included medical occurrence that, at any dose, results in death, is life threatening, requires hospitalization or prolongation of existing hospitalization, results in persistent or significant disability/incapacity, is a congenital anomaly/birth defect, is a significant medical event in the investigator's judgment, or is an event of possible drug-induced liver injury with hyperbilirubinaemia. SAEs were includes if the onset date was on or after the treatment start date and on or before the treatment stop date. However, the window for an SAE of pneumonia was longer and included 28 days post study treatment stop date.

A serious adverse drug reactions (SADR) is any untoward medical occurrence suspected to be medicinal product-related that at any dose: results in death, is life-threatening, requires inpatient hospitalization or prolongation of existing hospitalization, results in persistent or significant disability/incapacity, or is a congenital anomaly/birth defect.