Real-life use of fluticasone propionate/salmeterol in patients with chronic obstructive pulmonary disease: a French observational study

The economic and social burden of chronic obstructive pulmonary disease (COPD) is considerable, with a projection that COPD will be the third leading cause of death by 2020 [1]. Pharmacologic management for patients suffering from COPD is based on the use of bronchodilators, associated or not with a corticosteroid [2]. The therapy prescribed depends on COPD severity, prior treatment(s), their efficacy and tolerance, and patient preference regarding inhalers. In the most recent GOLD (Global initiative on Obstructive Lung Disease) global strategy document, four patient categories are defined (A, B, C and D) based on exacerbation risk and symptoms burden [2]. Exacerbation risk is assessed according to GOLD category for forced expiratory volume in 1 second (FEV₁) and past history of exacerbations, while symptoms burden is assessed by the modified Medical Research Council (mMRC) dyspnea grading scale and/or the COPD Assessment Test (CAT), based on the worst case scenario. For patients with FEV₁ < 50% predicted normal or a history of at least two exacerbations per year or one hospitalization for COPD exacerbation or both, combinations of inhaled corticosteroids (ICS) and long-acting β2 agonists (LABA) are recommend among first-line treatments.

Administration of the ICS fluticasone propionate with the LABA salmeterol xinafoate (fluticasone /salmeterol combination, FSC) with a dry powder inhaler was approved in Europe in 2003, at a dose of 500 μg fluticasone propionate with 50 μg salmeterol twice daily for treatment of COPD patients with a pre-bronchodilator FEV₁ < 60% predicted normal, a history of repeated exacerbations, and who have significant symptoms despite regular bronchodilator therapy. The change of the FEV₁ threshold from <50% in the GOLD guidelines [2] to <60% was mainly influenced by the TORCH study [3].

Coherence between published COPD guidelines and clinical prescription practices in the real-world setting has been brought into question [4–6]. Several reports highlight that a high proportion of patients who are prescribed inhaled medications, have not undergone spirometric testing to confirm diagnosis and severity [7–11]. Data on FSC use in the real-life COPD setting are scarce and co-therapies in patients prescribed with this combination as long-term treatment have not been reported.

Following a request by the French Health Authorities, a national, prospective, population-based, observational study was performed in adult COPD patients initiating treatment with fluticasone propionate/salmeterol (500/50 μg) administered with a dry powder inhaler, prescribed by a general practitioner (GP) or pulmonologist. The study purpose was twofold; the first step was to collect data describing patient characteristics and treatment modalities in this population, and to analyze the coherence of the prescription population with the licensed COPD population according to the Summary of Product Characteristics (SPC). In the second step, clinical outcomes after 12-months treatment with SFC in this population were evaluated. The present article reports results regarding the first of these two goals.

Knowledge of real-life use of ICS/LABA combinations in the COPD setting allows for evaluation of the conformity of current prescription practices with the licensed COPD population. This can in turn be used to encourage appropriate changes in current practices to improve standards of patient care. Limited data on the use of FSC in the real-life setting have been published since the addition of COPD to the original marketing authorization for this combination based on three double-blind randomized placebo-controlled studies [16–18].

From the data obtained in this observational study of a COPD population initiating treatment with FSC administered via a dry powder inhaler, the very low rates of strict compliance with the three SPC criteria for prescription, notably in ICS-naïve patients, clearly reveal that prescribing French physicians do not adequately respect the licensing conditions. Less than a quarter of the patients with prior ICS and less than 7% of ICS-naïve patients fulfilled all three conditions. Furthermore, a small proportion of patients did not fulfill any of the conditions. When considering both two or three of the conditions, conformance improves slightly, but is still low, being reported in approximately two-thirds of patients with prior ICS, and one-third of ICS-naïve patients.

While conformance was higher for pulmonologists than GPs in terms of the number of conditions respected and for the FEV₁ and previous bronchodilator criteria, overall rates were low for both medical specialties. In the case of GPs, the discrepancy between the licensed and real-life COPD populations is due at least in part to the high rate of missing spirometry measurements, with FEV₁ data available for less than half of these patients. This is coherent with several studies worldwide reporting that only around a third to a half of patients undergo spirometry testing for newly diagnosed COPD or subsequently during follow-up [7, 10, 11, 19, 20]. The large proportion of missing data for FEV₁ among GP patients limits the assessment of concordance between prescription practices and marketing authorizations and guidelines. The conservative approach used with missing data (i.e., absence of FEV₁ considered non-respect) may have artificially lowered the rate of conformance in the GP group. Supporting this, a sensitivity analysis including only patients for whom FEV₁ measures were available gave similar conformance rates between GPs and pulmonologists for this criterion (Additional file 4). Of note, the profile of patients without FEV₁ measures showed them to be generally healthier than those with measures (younger, less severe COPD and dyspnea, and better QoL; see Additional file 2).

It could also be hypothesized that perceived QoL is a particularly important determinant of treatment decisions by GPs. Indeed, health-related QoL as measured by the CCQ was similar in pulmonologist and GP patients despite the latter patients having a lower dyspnea grade, less pronounced airflow obstruction and less frequent hospital visits. This illustrates that QoL measures, even when specifically designed for COPD patients, capture the impact of components other than those directly related to COPD. Interestingly, GP patients were younger and less likely to be retired than those treated by pulmonologists. Thus, it could be hypothesized that their perception of the disease’s impact is enhanced by their activity requirements, although this possibility remains to be tested. It was also noted that although pulmonologist patients are likely to have more severe COPD, pulmonologists tend to be more restrictive in the use of antibiotics and corticosteroids in patients treated in the community, as recommended in the French COPD guidelines. In these guidelines, antibiotics are recommended only when sputum is purulent, and corticosteroids are to be prescribed only for patients with severe baseline airflow obstruction and/or lack of improvement following treatment with antibiotics, if required, and bronchodilators.

Another potential reason for discrepancy between the licensed and real-life COPD populations may lie with the possibility that exacerbations were underestimated in this study since they were only assessed from patient recall; interestingly, conformance with this criterion was particularly low among pulmonologists. A low level of overall conformity was maintained in a sensitivity analysis excluding this criterion. Difficulty evaluating prior exacerbations stems from multiple sources; not only is there an absence of consensus on the definition of COPD exacerbations [21–23], but in addition, patients tend to under-report exacerbations [24], even those that have clinical significance [25].

Differences in conformity were seen according to prior corticosteroid intake, with patients having prior ICS generally less likely to conform to prescription recommendations relative to ICS-naïve patients. This may have been influenced by the fact that prior ICS treatment may have modified their clinical profile [26]. It may also be explained by the fact that this population was older, had more severe and longer duration COPD, with a higher incidence of oral corticosteroid and antibiotic intake.

Poor physician compliance with licensed conditions may also reflect a level of difficulty in applying recommendations and guidelines in the face of lack of clarity or simplicity, or in the context of individualized patient care. As reported by Corrado et al., recommendations can be considered inappropriate if other factors such as pulmonary hyperinflation, exercise capacity and tolerance, or comorbidities are not taken into account [27]. This may also explain the relatively high rates of incorrect dose prescription. Dialogue with physicians, along with improved awareness and education are needed in order to address these issues.

Analysis of populations and SPC conformance according to the GOLD 2011 classification reveals some differences compared to GOLD 2006 classification. Very severe airflow obstruction (GOLD grade 4) was reported in approximately one-quarter of the population while 40% belonged to the D GOLD 2011 category. In nearly a quarter of cases, GPs prescribed FSC to patients with GOLD 2006-defined mild COPD. This may be due to the fact that they (along with pulmonologists) tended to underestimate COPD severity relative to both 2006 and 2011 GOLD classifications (notably for most severe disease). Comparing the GOLD 2006 and 2011 classifications also shows that according to the latter, GPs were more likely than pulmonologists to see patients with more severe COPD (74% versus 55%, respectively), which was not the case with the 2006 guidelines (32% versus 38%, respectively). Furthermore, according to the GOLD 2011 classification, more patients have more severe disease, suggesting a higher rate of conformity with the SPC licensing conditions, and also that the guidelines may be evolving to adapt to the real-life situation.

Another potential study limitation concerns the representativeness of participating physicians. The low rate of physicians selected randomly from the national population who agreed to participate in the study (<10%) suggests a general reluctance to participate and consequently a potential selection bias. Nonetheless, participating physicians were broadly representative of national figures for their respective specialties in terms of age, gender and practice settings [28], although male physicians were over-represented relative to national figures. It could be hypothesized that participating physicians were more likely to be interested in the field of COPD and prescribe treatments more adequately than less interested physicians. As a consequence, the already very high rate of non-concordance between prescriptions and guidelines could be under-estimated. However, we have no means of testing this pessimistic hypothesis.

The COPD population prescribed with FSC by both GPs and pulmonologists conforms poorly with the licensed population with less than a quarter of ICS-treated patients and 7% of ICS-naïve patients fulfilling all three marketing conditions. Prescribing clinicians need further education on the importance of following the SPC recommendations, including FEV₁ evaluation, to ensure FSC is used only in COPD patients who are likely to benefit from it. Understanding the reasons behind clinicians’ decisions to prescribe FSC to non-conforming patients may help to resolve this issue.